# Digital multi-iterative filter

FIELD: information technology.

SUBSTANCE: filter comprises difference units, correction units, adder units, delay circuits, matrix function units, shaping and readout unit of apriori data, regularisation parameter unit.

EFFECT: accuracy of estimating information process parameters in measuring systems.

7 dwg

The invention relates to digital computing and can be used in systems digital signal processing when solving problems of optimal nonlinear filtering.

A device - extended Kalman filter [1, 2], the lack of which is the limited functionality due to the linear structure of the processed processes and devices [3], and heuristic setting of the regularization parameter.

The closest to the technical nature of the claimed invention is a digital intelligent iterative filter [3], containing the first, second, and third blocks forming the difference; first, second, third correction blocks; block forming and outputting a priori data; first, second, third blocks forming amount; first, second, third, fourth, fifth and sixth blocks the formation of matrix functions; first, second and third delay lines; first and second block of the calculation of the regularization parameter. However, this device in some practically important cases does not allow to obtain the required accuracy.

Improving the accuracy characteristics of the filter parameters of random processes is relevant direction.

The claimed invention is directed to improving the accuracy of their estimates of the parameters of information is the process in measuring systems which is very important for radar tracking of targets containing blocks: first, second, third, fourth blocks forming the difference; first, second, third, fourth correction blocks; block forming and outputting a priori data; first, second, third, fourth blocks forming amount; first, second, third, fourth, fifth, sixth, seventh and eighth blocks the formation of matrix functions; first, second, third and fourth delay lines; first, second and third block of the calculation of the regularization parameter, the first, second, third and fourth outputs block forming and outputting a priori data are connected respectively with the second, third, fourth and fifth information inputs of the first, second, third and fourth correction block, in addition, the first, second, fifth outputs of the block forming and outputting a priori data are connected respectively with the third, second and fifth information inputs of the first, second and third block of the calculation of the regularization parameter, the first data output of the first correction block is connected with the first information input of the first shaping unit amount, the output of which is connected to the information input of the first shaping unit matrix functions and fourth information input of the first block of the calculation of the regularization parameter, first the information is hydrated input which is the input device; the output of the first block of the calculation of the regularization parameter is connected with the ninth informational input of the second error correction block, the output of the first processing unit matrix functions connected with the information input of the first delay line, the output of which is connected with the second information input of the first shaping unit amounts, with the seventh information input of the first error correction block and the information input of the second shaping unit matrix functions, the second information, the output of which is connected with the sixth information input of the first error correction block, the first data output of the second processing unit matrix functions connected with the second information input processing unit difference, the output of which is connected to the first information input of the first error correction block; the second information output of the first correction block is connected with the eighth informational input of the second error correction block, the first information output of which is connected to the first information input of the second processing unit amount, the output of which is connected to the information input of the third block the formation of matrix functions and with the fourth information input of the second block of the calculation of the regularization parameter, the output of which is connected to the ninth informational input of the third error correction block, the output of the third block formed the Finance matrix functions connected with the information input of the second delay line, the output of which is connected with the second information input of the second processing unit amount, the seventh information input of the second error correction block and the information input of the fourth processing unit matrix functions, the second information, the output of which is connected with the sixth information input of the second error correction block, the first data output of the fourth processing unit matrix functions connected with the second information input of the second processing unit difference, the output of which is connected to the first information input of the second error correction block; a second information output of the second error correction block is connected with the eighth informational input of the third error correction block, the first information output of which is connected to the first information input of the third processing unit sums the output of which is connected to the information input of the fifth processing unit matrix functions, the output of which is connected to the information input of the third delay line, the output of which is connected with the second information input of the third unit forming the sum, the seventh information input of the third block of the correction and the information input of the sixth processing unit matrix functions, the second information, the output of which is connected with the sixth information input to the third error correction block, the first info is to boost the output of the sixth processing unit matrix functions connected with the second information input of the third processing unit difference, the output of which is connected to the first information input of the third error correction block; a second information output of the third error correction block is connected with the eighth informational input of the fourth correction block, the first information output of which is connected to the first information input of the fourth processing unit amount, the output of which is an output device connected to the information input of the seventh processing unit matrix functions, the output of which is connected to the information input of the fourth delay line, the output of which is connected with the second information input of the fourth processing unit amount, the seventh information input of the fourth block of the correction and the information input of the eighth block the formation of matrix functions, the second information, the output of which is connected with the sixth information input of the fourth correction block, the first data output of the eighth block the formation of matrix functions connected with the second information input of the fourth processing unit difference, the output of which is connected to the first information input of the fourth correction block.

One way to improve the accuracy of parameter estimation of dynamic systems is the use of methods of solving ill-posed problems based on the principles of regularization. The effectiveness of R is polarizatio for the case of continuous systems proven and received its expression in the form of A. N. Tihonov's method [4] and its variants in the form of a method of iterative regularization [5]. Get the ltration equation using the method of iterative regularization for discrete systems [6-9].

Let the dynamics of the estimated parameters is described by the system of difference equations in discrete-time

wherethe state vector of the system under investigation;the vector of unknown external influences; a transition function- continuous together with partial derivatives of the vector-function of its arguments; (G∈E^{M}×E^{M}matrix of the intensity of external influences; k, N, M be positive integers. It is assumed that the matrixhas an inverse matrix.

The observed signal obtained at the output model of the measurement system, described by a discrete equation

wherethe vector of observations

the vector of discrete white Gaussian noise with known local characteristics

W - covariance matrix of dimension L×L, δ(·) is a vector of Delta - function;signal vector function, continuous together with partial derivatives; L, l be positive integers.

Pose the problem of synthesis recurren the aqueous filter evaluation x^{
*}(k), optimal in the sense of minimum functionality that characterize the measurement error

In virtue of the continuity of the vector - function F(·), the solution of equation (1) continuously depends on η(k), therefore the error function (3) for each solution of system (1) continuously depends on η(k). Thus, the problem of determining the estimate x^{*}(k)delivering a minimum of (3)is equal to the task definition

Task(1), (2), (4) is an ill-posed inverse problem [7]. Find the values of vectors x^{*}(k), η^{*}(k)by solving the set of equations(1), (2), (4) in terms of the incorrectness of the original problem is quite difficult, therefore, widespread iterative gradient methods. However, the use of such methods may lead to divergent sequence of approximations. Therefore, the use of any iterative method for solving the problem(1), (2), (4) requires the definition of a regularizing family of operators, in which the regularization parameter is the number of the iteration.

In accordance with the General definition of regularizing family of operators Antigonos [7], we say that the iterative method

in which a numeric parameter α_{n}satisfies the conditions is m:

where Δ(η_{n}) - the discrepancy generates regularizing family of operators, in which the parameter is the number of the iteration, if for any initial approximation η_{0}and for any value of the error of the initial data σ, satisfying the condition of 0<|σ| < σ_{0}, σ_{0}=const, there is a number n(σ), such that

that is, the obtained approximations converge to the exact solution in the norm spacewhen aiming error of the original data to zero.

According to [8] the expression to determine the gradient at the point η_{n}(k) has the form

where x_{n}(k) is the solution of (1) when η_{n}(k), and the vector ψ_{n}(k) is determined from conditions

Knowing the expression for the gradient (7) the functional (3), we can rewrite expression (6) for the regularization parameter [9] in the following form

The selection of the sequence of parameters α_{n}satisfying the condition (9), will implement the digital filter processing of measuring information increased accuracy.

For the implementation of the iterative method (5) is required to determine the gradient of the functional (3)defined by the expression (7). Assuming zero approximation η_{0}(k)=0,write the iterative sequence (5) in the expanded form for

As a result, taking into account (7), we have a sequence of discrete two-point boundary value problems (DTCS)

Let us introduce the notationand panajim each of the equations for the adjoint vector ψ_{i}the value α_{i}then equation (10) take the following form

To obtain the recurrent algorithm of estimation of the state vector, you must use the method of invariant embedding in discrete form. Note that the equation for the vector functions λ in DTCS (9) written in reverse time. This requires converting it to a form reflecting the dependence of λ_{n}(k+1) λ_{n}(k) and x_{n}(k). Making the proper conversions taking into account the expressions for x_{n}(k+1) from (11) and using the decomposition in a Taylor series in a neighborhood of F(x_{n}(k),k), we obtain the following sequence DTCS

where the functions β and γ are introduced to reduce the entry.

We replace the condition at the end of λ_{n}(N)=0 a more General condition λ_{n}(N)=and let N and C - variables. Then the value of the vector x_{n}(N) is determined as a function of N and with

x_{n}(N)=r[s,N].

Change of size N Ha N+1 gives the increment Δ, then

Let us write the expression for r (+Δ, N+1), using the apparatus finite difference

r (+Δ, N+1)=r(c, N)+Δr_{c}(C, N)+Δr_{N}(C, N)+Δ^{2}r_{cN}(C, N),

or, given (13), we obtain

where

According to (12) the expression for Δx_{n}and Δ have

To solve the differential equation (14) with respect to r(c,N), i.e. to find a General analytical solution is not possible, and usually turn to approximate methods. Suppose that r(c,N) is linear with

where- assessment of the state vector at time N, P_{n}(N) is some matrix with dimension M×M

Calculate the difference included in the expression (14)using the expression (16)

Substituting expressions(15), (16), (17) in (14), we obtain

Decomposing β and γ in a Taylor series in the neighborhoodand neglecting terms of order higher than the first, we can write equation (18) in the form

Equation (19) is performed when→0, therefore, by equating the coefficients of the first and zero degree, obtain a differential equation forand

Write DTCS (12) for the case when k=N it should be noted that this is all DTCS for i=0,...,n-1 are resolved and consequently the evaluation of x_{i}are known functions of the parameter k. Have

Because of the difference in Perevoznaya matrices P_{n}andno, we record the sequence of equations for estimatingprocess (1), assuming that N is constantly changing and k=N, and taking into account the condition (7) is imposed on the regularization parameter, in the form

The sequence of equations (24) is a digital intelligent multi-stage filter, which allows to process the digital processing of the measurement data for discrete dynamical systems. If we compare the equations with the equations of the recursive digital filter [3], it becomes clear that they are different from each other due to the introduction of additional iterations, as well as additional links sequence parameters α_{i}changing the total feedback factor in the equation to estimatewith estimatesthe signal vector function H and the matrices G, W. That is, in the filter (24), in contrast to the filter [3], the parameter regularis the tion agreed in accordance with the discrepancy principle with measurement errors,
which allows to obtain a more accurate assessment procedure. The algorithm (24) gives the optimal nonlinear system parameter estimation in the sense of minimum of the functional characterizing the RMS error of the measuring channel.

The calculation of the regularization parameter is organized as follows: the limits of integration [0,t] is changed to [t-4s, t], where t denotes the current time, the s - step calculations; for discrete time you must take [N-1, N-5]. Studies have shown [6]that the use of data more than four steps back provides increased accuracy of the resulting estimates by a fraction of a percent when the increase in the number of necessary arithmetic operations several times. A method that uses more than four iterations will be called we.

It should be noted that for the implementation of the 2nd iteration of the algorithm (24), you need to 1st iteration was implemented, for the implementation of the 3rd - 1st and 2nd for the implementation of the 4th - 1st, 2nd and 3rd. That is, the number of arithmetic operations required to compute the iterations of the algorithm will include the number of arithmetic operations of the previous iterations. Analysis of the computational cost required for the implementation of the developed algorithm estimates allow us to conclude about the possibility of its implementation in real-time based on the use of district is authorized computing.

Studies have shown that the effectiveness of each re-processing with respect to the previous falls. Meanwhile, the amount of computation increases dramatically. Based on the analysis of the results of the numerical simulation can be recommended four-iterative data processing.

The evaluation of the effectiveness of the designed filter is produced on the basis of numerical simulation of the problem of determining the unknown constant parameter d discrete nonlinear system of the third order

where the parameter τ has the meaning of a time interval which receives the measurement information in the form

Thus, as follows from relations (24), the introduction of new structural elements and connections allows in combination with the overall characteristics to obtain the technical result consists in the reduction of the dispersion error is obtained at the output of the filter estimates of the input processes.

The device relates to digital computing and can be used in systems of digital processing of radio signals for solving problems of optimal nonlinear filtering, as well as for detecting maneuvering targets, as it is very important both from a tactical and technical point of view. Technically intensive maneuvering the ate can cause a significant degradation in the accuracy of the sustainability of its auto-tracking servo gauges. In connection with the above circumstances in the composition of modern and advanced converters it is advisable to include a special device (algorithms) for detection, tracking, targeting, etc. and the correction (adaption) of the parameters or structure of filters in accordance with this setting.

The disadvantage of the selected prototype is a low-precision generated estimates of the information process. Using the calculation of the regularization parameter according to the new formula in this paper improves the accuracy of the estimation of the parameters of this device.

The calculated weight of the regularization parameter has the form [10]:

On the basis of the prototype below is the description of the invention is calculated by the formula (27).

The invention is illustrated in figure 1-7, which shows the structural diagram of the digital we filter the first and second correction blocks, block calculation accuracy characteristics, the unit of calculation of the regularization parameter, the processing unit works the numerator of the regularization parameter, unit calculations.

Figure 1 shows the structural diagram of the digital we filter. The device includes a first block 1, the second block 7, the third block 14 and the fourth block 20 forming the difference; the first block 2, the second block of 8, the third block 15 and h is twenty unit 21 correction; the first block 3, the second block 9, the third block 16 and the fourth block 22 forming the sum of the first delay line 5, the second delay line 11, a third delay line 18 and the fourth delay line 24; the first block 4, the second block 6, the third block 10, fourth block 12, the fifth unit 17, the sixth block 19, the seventh block 23 and the eighth block 25 of the formation matrix functions, block 13 forming and outputting a priori data; the first block 26, the second block 27 and the third block 28 calculation of the regularization parameter.

Figure 2 shows the structural diagram of the first error correction block, which contains the block 2.1 formation of partial derivatives, unit 2.2 transpose matrix functions, block 2.3 formation works, unit 2.4 calculation accuracy characteristics, block 2.5 formation works.

Figure 3 shows the structural diagram of the second error correction block, which contains the block 8.1 formation of partial derivatives unit 8.2 transpose matrix functions, block 8.3 formation works, block 8.4 formation of the sum block 8.5 calculation accuracy characteristics, block 8.6 formation works.

Figure 4 shows a structural block circuit diagram of the calculation accuracy characteristics included in the first, second, third and fourth blocks of corrections, which contains the block 20 forming the partial derivatives of matrix functions, block 21 transport the financing of matrices, block 22 forming compositions, line 23 of the delay unit 24 transpose of matrices, block 25 forming compositions, the block 26 forming the sum, unit 27 of the formation of partial derivatives of matrix functions, block 28 forming compositions, the block 29 forming the difference, the device 30 of conversion matrices, the block 31 of the formation of the work.

Figure 5 shows a structural block circuit diagram of the calculation of the regularization parameter, which contains the set of matrix functions 26.1, delay lines 26.2, 26.4, 26.8, 26.9, 26.13, 26.14, 26.18, 26.19, 26.20, 26.26, at 26.27, 26.28, at 26.32, 26.33, 26.34, 26.35, 26.43, 26.44, 26.45, 26.46, blocks 26.3, 26.10, 26.12, 26.21, 26.24, 26.25, at 26.36, 26.40, 26.41, 26.42 forming compositions of the numerator of the regularization parameter, blocks 26.5, 26.15, at 26.29, 26.47 transpose blocks 26.6, 26.16, 26.30, 26.48 formation works, blocks, 26.7, at 26.17, 26.31, 26.49 forming the difference, blocks 26.11, 26.22, 26.23, 26.37, 26.38, 26.39 calculations, blocks 26.51, 26.50 forming sums, unit 26.52 formation works denominator of the regularization parameter, the block 26.53 form relationships.

Figure 6 shows the structural diagram of the first processing unit works the numerator of the regularization parameter included in the first, second and third forming unit of the regularization parameter, which contains the block 26.3.1 the formation of matrix functions, block 26.3.2 partial derivatives of matrix functions, BL is to 26.3.3 transpose of matrices, block 26.3.4 difference, block 26.3.5 formation works, blocks 26.3.6 and 26.3.7 transpose matrices and block 26.3.8 formation works.

Figure 7 shows the structural diagram of the first block of calculation of the forecast, which is part of the shaping unit works the numerator of the regularization parameter, it includes blocks 26.11.1 and 26.11.5 formation works, blocks 26.11.2 and 26.11.6 forming sums, unit 26.11.3 the formation of the partial derivatives of matrix functions, block 26.11.4 the formation of matrix functions.

For the first block of the calculation of the regularization parameter (figure 5) informational inputs of the delay lines 26.2, 26.8, 26.18, at 26.32, connected with the information output unit 3 forming amount (figure 1) the first informational inputs of the delay lines 26.4, 26.13, 26.26, 26.43, connected with the first information output from the first block 26.1 forming the matrix functionsinformation input connected to the information output unit 3, the information output of the first delay line 26.2 connected with the second information input of the first block 26.3 forming compositions of the numerator of the regularization parameter, the first and the third information input is connected to the second and first information output unit 13 forming and outputting a priori data, respectively, and the fourth information input from the input device, the first information output from the first block 26.3 forming compositions of the numerator of the regularization parameter is connected with the sixth and seventh informational inputs of the second unit 26.50 formation amount of the first information, the output of which is connected to the first information input unit 26.53 forming relationship, the first output unit 26.53 forming relationships is the output of block 26; first information output of the second delay line 26.4 connected to the first information input of the first block transpose 26.5 and the first information input of the first block 26.6 formation works, the first information output unit 26.5 connected with the second information input unit 26.6 formation works, the first information output of which is connected with the second information input of the first block 26.7 forming the difference between the first information input of which is connected to the fourth information input unit 26.3, the first information output unit 26.7 connected with the fourth information input of the first block 26.51 formation amount of the first information, the output of which is connected to the first information input unit 26.52 formation works denominator of the regularization parameter and the second input unit 26.52 connected to the fifth output unit 13 forming and outputting a priori data; the first informationdavid block 26.52 formation works denominator of the regularization parameter is connected with the second input unit 26.53 forming relationships; the first information output thirds 26.8 delay line connected to the first information input of the fourth 26.9 delay line, the first information output of which is connected with the second information input of the second block 26.10 forming compositions of the numerator of the regularization parameter and the second information input of the first block 26.11 calculation of the forecast, the second information, the output of which is connected with the second information input of the third block 26.12 shaping the work of the numerator of the regularization parameter, the third and fourth information input blocks 26.10, 26.12, connected with the first information output unit 13 and the input device, the first information output unit 26.10 connected with the fourth and third information input unit 26.50, the first information the output of block 26.12 connected with the tenth information input unit 26.50 formation amount; the first data output of the fifth 26.13 delay line connected to the first information input of the sixth 26.14 delay line, the first information output of which is connected to the first information input of the second block 26.15 transpose and the first information input of the second block 26.16 formation works, the first data output of the second 26.15 block transpose is connected with the second information input unit 26.16, the first information you the od which is connected with the second information input of the second processing unit difference at 26.17, the first information input of which is connected to the fourth information input unit 26.10 and the first information output unit at 26.17 connected with the third information input unit 26.51 formation amount; the first information output of the seventh delay line 26.18 connected to the first information input of the eighth delay line 26.19, the first information output of which is connected to the first information input of the ninth 26.20 delay line, the first information output of which is connected with the second information input of the fourth 26.21 block forming compositions of the numerator of the regularization parameter and the second unit 26.22 calculation of the forecast, the first information output of which is connected with the first information input of the third block 26.23 calculation of forecast and 26.25 formation works the numerator of the regularization parameter, the third and fourth information input blocks 26.21, 26.24 and 26.25 connected with the first information output unit 13 and the input device, the first information output unit 26.21 connected with the twelfth and eleventh informational inputs of the second unit 26.50 forming amount, the first information outputs blocks 26.25, 26.24 connected with the thirteenth and fourteenth informational inputs of the first block 26.50 respectively; the first information output of the tenth delay line 26.26 connected to the PE the new information input eleventh delay line at 26.27, the first information output of which is connected to the first information input of the twelfth 26.28 delay line, the first information output of which is connected with the first information input of the third block at 26.29 transpose matrix functions and the first information input of the third block 26.30 formation works, the first data output of the third unit at 26.29 transpose is connected with the second information input of the third block 26.30 formation works, the first information output of which is connected to the first information input of the third processing unit difference 26.31, the first information output of which is connected to the first information input unit 26.51 forming amount, and the first information input of the third block 26.31 connected with the fourth information input unit 26.21; the first information output thirteenth delay line at 26.32 connected to the first information input of the fourth delay line 26.33, the first information output of which is connected to the first information input of the fifteenth 26.34 delay line, the first information output which in turn is connected to the first information input of the sixteenth delay line 26.35, the output of which is connected with the second information input of the seventh unit at 26.36 forming compositions of the numerator parameter regulari the emission and the second information input of the fourth block 26.37 calculate forecast the second information output of which is connected with the second information input of the ninth block 26.41 forming compositions of the numerator of the regularization parameter and the second information input of the fifth block 26.38 calculation of the forecast, the second information output of the latter is connected with the second information input of the tenth block 26.42 forming compositions of the numerator of the regularization parameter and the second information input of the sixth computing unit forecast 26.39, the second information, the output of which is connected with the second information input of the eighth block 26.40 forming compositions of the numerator of the regularization parameter, the third and fourth information input blocks at 26.36, 26.40, 26.41, 26.42 connected respectively with the first information output unit 13 and the input device, the first information the output of the block at 26.36 connected to the ninth and eighth information input unit 26.50 forming amount, the fourth input of this block is connected to the first input of the fourth processing unit difference 26.49; the first information output unit 26.40 forming compositions of the numerator of the regularization parameter is connected with the fifth information input unit 26.50, the first information outputs blocks 26.41 and 26.42 forming compositions of the numerator of the regularization parameter is connected with the first and second information inputs the Loka 26.50 respectively; the first information output seventeenth delay line 26.43 connected to the first information input of the eighteenth delay line 26.44, the first information output of which is connected to the first information input of the nineteenth 26.45 delay line, the first information output which in turn is connected to the first information input of the twentieth delay line 26.46, the information output of which is connected with the first information input of the fourth block 26.47 transpose and fourth block 26.48 formation works, the first information output unit 26.47 connected with the second information input unit 26.48, the first information output of which is connected with the second information input unit forming asnosnosci 26.49, which is connected with the second input information block 26.51 formation amount; and the first information input unit 26.49 connected with the fourth information input unit at 26.36; the first information input of the second 26.10, fourth 26.21 and seventh at 26.36 blocks forming compositions of the numerator of the regularization parameter is connected with the second information output unit 13 forming and outputting a priori data and the first information input of the third 26.12, fifth 26.24, sixth 26.25, eighth 26.40, ninth 26.41 and tenth 26.42 blocks forming compositions of the numerator parameter is the regularization connected, accordingly, with the first informational outputs of the first 26.21, third 26.23, second 26.22, sixth 26.39, fourth 26.37 and fifth 26.38 blocks calculation of the forecast; the first information input of the first 26.11, the second 26.22 and fourth 26.37 blocks calculate the forecast is connected to the second information output unit 13, and the first information input of the third 26.23, fifth 26.38, sixth 26.39 blocks calculate the forecast connected with the first information outputs, respectively, of the second 26.22, fourth 26.37 and fifth 26.38 blocks calculation of the forecast, the third information input blocks calculate the forecast is connected with the fifth information output unit 13. The structure and operation of the second and third block of the calculation of the regularization parameter is similar.

First, second, and fifth information output unit 13 forming and outputting a priori data (figure 1) are connected respectively to the third, second and fifth information inputs of the first block 26, the second block 27 and the third unit 28 calculation of the regularization parameter, the output of which is connected to the ninth informational input of the fourth correction block 21, and the first, second, third, fourth information output unit 13 forming and outputting a priori data are connected respectively with the second, third, fourth, fifth information inputs of the first block 2, the second unit 8, t is its block 15 and the fourth block 21 correction, the second information output of the first correction unit 2 is connected with the eighth informational input of the second correction unit 8, the second information output of the second correction unit 8 is connected with the eighth informational input of the third correction unit 15, the second information output of the third correction unit 15 is connected with the eighth informational input of the fourth correction block 21, the output of which is connected to the first information input of the fourth block 22 forming amount, the output of which is the output of the device, and also connected to the information input of the seventh block 23 forming the matrix functions whose output is connected to the information input of the fourth delay line 24, the output of which is connected with the second information the fourth input unit 22 forming the sum, with the seventh information input of the fourth block 21 of the correction and the information input of the eighth block 25 forming the matrix functions, the second information, the output of which is connected with the sixth information input of the fourth correction block 21; the first information output of the eighth block 25 forming the matrix functions connected with the second information input of the fourth block 20 forming the difference, the output of which is connected to the first information input of the fourth correction block 21; the output of the first block 26 calculation of the pair is of ETP regularization is connected with the ninth informational input of the second correction unit 8, the output of which is connected to the first information input of the second unit 9 forming amount, the output of which is connected with the fourth information input of the second block 27 calculation of the regularization parameter and the information input of the third block 10 forming the matrix functions whose output is connected to the information input of the second delay line 11, the input of which is connected with the second information input of the second unit 9 forming the sum, with the seventh information input of the second correction unit 8 and with the information input of the fourth block 12 forming the matrix functions, the second information, the output of which is connected with the sixth information input unit 8 correction; the first data output of the fourth block 12 forming the matrix functions connected with the second information input of the second unit 7 forming the difference, the output of which is connected to the first information input of the second correction unit 8; the first information input of the first correction unit 2 is connected with the first information input of the first unit 3 forming amount, the output of which is connected with the fourth information input of the first block 14 calculation of the regularization parameter and the information input of the first block 4 forming the matrix functions whose output is connected to the information input of the first delay line 5, the output is toroi connected with the second information input of the first unit 3 forming the sum with the seventh information input of the first correction unit 2 and the information input of the second block 6 forming the matrix functions, the second information, the output of which is connected with the sixth information input of the first unit 2 correction; first information output of the second block 6 forming the matrix functions connected with the second information input unit 1 forming the difference, the output of which is connected to the first information input of the first correction unit 2; the first information input of the first block 1 forming the difference between the first information input of the second unit 7 forming the difference between the first information input of the third block 26 forming the difference and the first information input of the fourth block 20 forming the difference, and also, the first information input of the first block 26 calculation of the regularization parameter, the first information input of the second block 27 calculation of the regularization parameter and the first information input of the third block 28 calculation of the regularization parameter are input devices.

The first and fourth information output unit 13 forming and outputting a priori data connected with the third and fourth information input unit 2.3 formation works (figure 2); the information output of the first block 1 forming the difference is connected with the first information photoblog 2.3 formation works; the second information output of the second block 6 forming the matrix functions connected with the information input unit 2.1 the formation of the partial derivatives, the output of which is connected with the information input unit 2.2 transpose matrix functions, the output of which is connected with the second information input unit 2.3, the output of which is connected to the first information input unit 2.4 calculation accuracy characteristics, the output of which is connected to the first information input unit 2.5 the formation of works whose output is the output of the first correction unit 2; an information output of the delay line 5 (figure 1) is connected with the second information input unit 2.4 calculation accuracy characteristics; the second and third outputs of the block 13 forming and outputting a priori data connected with the third and fourth information input unit 2.4 calculation accuracy characteristics; an output unit 2.3 formation of works connected with the second information input unit 2.5 formation works and the second information input unit 8.4 formation (figure 3).

The information output of the second unit 7 forming the difference is connected with the first information input unit 8.3 forming compositions (figure 3). The first information output unit 13 forming the issuance of a priori data is connected with the third information shall include the input unit 8.3 formation works; the output of block 26.53 connected with the second information input unit 8.3 formation works; the second information output of the fourth block 12 forming the matrix functions connected with the first information input unit 8.1 formation of partial derivatives, the output of which is connected with the information input unit 8.2 transpose matrix functions, the output of which is connected with the fourth information input unit 8.3, the output of which is connected to the first information input unit 8.4 formation amount of the second information input of which is connected to the information output unit 2.3 (figure 2), the output unit 8.4 formation of the amounts connected with the first information output unit 8.5, and with the second information the output unit 8.6 formation works; the third and fourth information output unit 13 forming the issuance of a priori data is connected with the second and third information input unit 8.5 calculation accuracy characteristics; information output of the delay line 11 is connected with the fourth information input unit 8.5 calculation accuracy characteristics, the output of which is connected to the first information input unit 8.6 formation works whose output is the output of the second correction unit 8 (figure 1).

The output of block 2.3 formation works (figure 2) is connected with the information the first input unit 27 of the formation of partial derivatives (figure 4), the output of which is connected with the information input unit 28 forming compositions, the output of which is connected to the first information input unit 29 of the formation of the difference, the output of which is connected with the information input device 30 accesses matrices, the output of which is connected to the first information input unit 31 of the formation of works whose output is the output of 2.4 (figure 2) calculation accuracy characteristics; the fourth output unit 13 forming and outputting a priori data is connected with the second information input unit 29 of the formation of the difference; the output of block 31 of the formation of works connected with the information input line 23 of the delay, the output of which is connected with the third information input unit 22 the formation of the works, the output of which is connected with the second information input unit 26 forming amount, the output of which is connected with the second information input units 28 and 31 forming compositions; the output of the first delay line 5 (figure 1) is connected with the information input unit 14 forming the partial derivatives, the output of which is connected with the information input unit 21 transpose matrices and the first information input unit 22 forming compositions, the second information input connected to the output of block 21; the third information output unit 13 Faure, the financing and the issuance of a priori data is connected with the information input unit 24 transpose matrices and the second the information input unit 25 formation works the first information input of which is connected to the information output unit 24; the output unit 25 is connected with the information input unit 26 forming amount.

Unit 8.5 calculation accuracy characteristics (figure 3) has a structure and communication, similar to the 2.4 block.

The first and second information output unit 13 forming and outputting a priori data connected with the fourth information input unit 26.3.5 (figure 6) and an information input unit 26.3.6 transpose matrix functions, the information output of which is connected with the third information input unit 26.3.5, the information output of which is connected with the information input unit 26.3.7 transpose matrix functions and the second information input unit 26.3.8 formation works, information output which is the output of the first shaping unit works the numerator of the regularization parameter, the information output unit 26.3.7 connected to the first information input unit 26.3.8; the first information input unit 26.3.5 connected with the information input unit 26.3.4 the second information input unit 26.3.5 connected with the information output unit 26.3.3 transpose of a matrix function, an information input connected to the information output unit 26.3.2 the formation of the partial derivatives of matrix functions, the information output is otorongo, in turn, is connected with the information input processing unit matrix functions; structure of all blocks forming the product of the numerator of the regularization parameter are identical, they only differ in the relationships between information input blocks 26.3.1 (connected either with the information output of the corresponding delay line, or the first information output unit of calculation of the forecast), and block 26.3.6 (information input connected either with the second information output unit 13 forming and outputting a priori data or the second information output unit calculations).

The first information input processing unit 26.11.1 works, which is the first information input of the first block of calculation of the forecast (figure 7) is connected with the information output of the delay line, the second information input unit 26.11.1 connected with the fifth information output unit 13 forming and outputting a priori data, with the same information output connected to the first information input unit 26.11.2 formation amount and a second information output unit 26.11.5, third information input unit 26.11.1 connected with the information output unit 26.11.3 the formation of the partial derivatives of matrix functions, the second information input of which is connected with the fifth information input unit 13 forming and revealing the Chi a priori data, and the first information input - information first access unit 26.11.4 the formation of matrix functions, the second information, the output of which is connected to the first information input unit 26.11.5 formation works and the second information input unit 26.11.6 formation the amount of information output unit 26.11.5 connected to the first information input unit 26.11.6; an information output unit 26.11.1 connected with the second information input unit 26.11.2, its informational output is the second information output unit calculations and is connected with the third information input unit 26.3.6 the second shaping unit works the numerator of the regularization parameter (figure 6); the structure of the subsequent blocks calculations similar, differ only connections between the third information input unit 26.3.5 (it is connected either with the second information input unit forecast, or with the fifth information output unit 13 forming and outputting a priori data and the first information input unit 26.3.1 (it is connected or with the information output of the delay line, or the first information output of the previous block of the forecast).

The device operates as follows (figure 1). In the initial state in block 13 forming and outputting a priori data recorded values of the matrices W^{-1}, G, I, and is also the value of α_{
0}, Δt is the value of the sampling step. The value of the information process in the (k+1)-th timefrom the output unit 3 forming the sum is fed to the input of block 26 computation of the regularization parameter and the input unit 4 forming the matrix functions, on the other inputs of the block 26 receives the values of y, G, W^{-1}with the output of block 4 forming the matrix functionsis fed to the input of the delay line 5, the output of which isis fed to the input unit 3 forming amount, the input correction unit 2 and the input unit 6 forming the matrix functionswhose value output unit 6 is fed to the input unit 2 and the input unit 1, the output of block 1 is the value ofresidual measurement, which is fed to the input of the correction unit 2, the other input of which receives the values of α_{0}, G, W^{-1}, I; in block 2 is formed by the product of the matrix of gain and residual measurements, which is summarized in unit 3 with a value of; the output of block 26, starting from the fourth point in time, a value of α_{1}of the regularization parameter (until then have α_{1}=0.5), which is fed to the input of the second correction unit 8, on the other inputs of the block 26 postepowania α_{
0}, G, W^{-1}, I; with one of the outputs of block 2 correction value M_{0}(k+1/k) to the input of the correction unit 8, which is formed by the value of

which is fed to the input of block 9 of the formation of the sum;the output unit 9 is fed to the input of block 27 calculation of the regularization parameter and the input unit 10 forming the matrix functions, from the output of block 10 isis fed to the input of the delay line 11, the output of which is formed isthat is summed with the value (27) in block 9, the output of which is formed isfrom the output of the block 11 isto the input of the correction unit 8, the input unit 9 and to the input unit 12 forming the matrix functions, the output of which is formed issupplied to the input unit 8 and the input unit 7, the other input of which receives an input oscillation; the discrepancy of measurementsfrom the output unit 7 is fed to the input unit 8, the other input of which receives the values of α_{1}, G, W^{-1}, I; with one of the outputs of block 8 of the correction value M_{1}(k+1/k) to the input of the correction unit 15, which is formed by the value of

is AutoRAE is fed to the input block 16 forming amount;
the output of block 16 is fed to the input unit 17 of the formation matrix functions, the output unit 17 is) is fed to the input of the delay line 18, the output of which is formed isthat is summed with the value (28) in block 16, the output of which is formed is; from the output of block 18 isto the input of the correction unit 15, the input unit 16 and to the input of block 19 of the formation matrix functions, the output of which is formed issupplied to the input unit 15 and the input unit 14, the other input of which receives an input oscillation; the discrepancy of measurementsfrom the output unit 14 is fed to the input unit 15, the other input of which receives the values of α_{2}, G, W^{-1}I, with one of the outputs of the block 15 correction value M_{2}(k+1/k) to the input of the correction block 21, which is formed by the value of

which is fed to the input unit 22 of the formation of the sum;the output of block 22 is fed to the input unit 23 of the formation matrix functions, from the output unit 23 isis fed to the input of the delay line 24, the output of which is formiruet the SJ
that is summed with the value of (29) in block 22, the output of which is formed is; output block 24 isto the input of the correction block 21, to the input unit 22 and the input unit 25 of the formation matrix functions, the output of which is formed iswhich is supplied to the input unit 21 and the input unit 20, the other input of which receives an input oscillation; the discrepancy of measurementsfrom the output of block 20 is fed to the input unit 21, the other input of which receives the values of α_{2}, G, W^{-1}, I.

The first correction unit 2 operates as follows (figure 2). The values of the matrix functionsgo to the input unit 2.1 the formation of the partial derivatives, with which the values ofgo on an input block transpose matrix functions 2.2, the output of whichthe discrepancy of measurementsand the values of W^{-1}α_{0}arrive at the inputs of the block 2.3 formation works, with which the value of M_{0}(k+1/k) is fed to the input of block 2.4 calculation accuracy characteristics, to the other input of which receives the values G, I,and the output of which is formed a P value_{1}(k+),
which is fed to the input of block 2.5 formation works, the other input of which receives the value of M_{0}(k+1/k) from the output of the block 2.3. The output of the 2.5 block is the output of the correction unit 2.

The second correction unit 8 operates as follows (figure 3). The value of the matrix functionsto the input of the block 8.1 formation of partial derivatives with which the values ofgo on an input block transpose matrix functions 8.2, with which the values ofand the residual value measurementsα_{1}, W^{-1}fed to the input of block 8.3 formation works, the output of which isto the input of the block 8.4 summing up, the other input of which receives the value of M_{0}(k+1/k); value

output unit 8.4 fed to the input of block 8.5 calculation accuracy characteristics, to the other input of which receives the values of G, W^{-1},and the output is formed by the value of P_{1}(k+1), which is multiplied by (28) in block 8.6 formation works; output block 8.6 is the output of the correction unit 8.

The first block 2.4 calculation of the accuracy characteristics of works in the following way (figure 4). Value a matrix options and
to the input of the block 20 forming the partial derivatives, the output of which isis fed to the input unit 21 of the transpose matrix functions and the input unit 22 forming compositions, the input of which also receives the value offrom the output of block 21 and a value of P_{0}(k) from the output of the delay line 23, at the entrance of which is from the output of block 31 forming the work, which is the output of the 2.4 block is a value of P_{0}(k+1);from the output of the unit 22 is fed to the input of block 26 forming the sum, the other input of which receives the value of GG^{T}formed in the block 25 of the formation of a work, on which input receives the value of G and the value of G^{T}formed in the block 24 of the transpose matrix, the input of which also receives the value of G; is the matrix functions M_{0}(k+1/k) is fed to the input of block 27 of the formation of partial derivatives, the output of which isto the input of the block 28, the other input of which receives a value of P_{0}(k+1/k)generated at the output of block 26;to the input of block 29 forming the difference, the other input of which receives the value of I; output unit 29 isis fed to the input condition the device 30 of conversion matrices,
the output of which is connected to the input unit 31 of the formation of the work, the other input of which receives a value of P_{0}(k+1/k) from the output unit 26. The unit of calculation accuracy characteristics 8.5 second correction block works the same way. The output of block 8.5 formed a value of P_{1}(k+1).

The unit of calculation of the regularization parameter works as follows. In the initial state to the input unit of the calculation of the regularization parameter goes- the value of the information process, where k is the current time, which, respectively, is transmitted to the delay line 26.2, 26.8, 26.18, at 26.32 (figure 5) and block the formation of matrix functionsfrom the output of which isis fed to the input 26.4, 26.13, 26.26 and 26.43 delay lines; in the next iteration of the filter on the same blocks receives the current value assessment of the information process, from the outputs of the delay lines 26.2, 26.8, 26.18, and at 26.32 26.4, 26.13, 26.26, 26.43 on delay line 26.9, 26.19, 26.33 and 26.14, at 26.27, 26.44 act accordingly k-1 the value of evaluation and matrix functionsthese values are estimates and matrix functions are received at the inputs of blocks 26.3 forming compositions of the numerator of the coefficient regularization and blocks 26.5 transpose matrix functions, 26.6 formation works; next on the moves of the delay lines 26.2,
26.8, 26.18, at 26.32 again enters the current value assessment of the information process, with these delay lines k-1 value assessment arrives at 26.9, 26.19, 26.33 delay line, the output of which k-2 is the assessment of the information process goes on to the input unit 26.10 and delay lines 26.20, 26.34; and similarly for and 26.4, 26.13, 26.26, 26.43, and 26.14, at 26.27, 26.44 delay lines, only for the matrix functions N to n+5 and subsequent iterations will have on the outputs of the delay lines 26.2, 26.9, 26.20, 26.35 values k-1, k-2, k-3, k-4 estimates of the information process, respectively, and outputs 26.4, 26.14, 26.28, 26.46 delay line values of the matrix functions for the same assessments. These values are estimates of the information process come into blocks 26.3, 26.10, 26.21, at 26.36 forming compositions of the numerator of the coefficient regularization and 26.11, 26.22, 26.37 generation forecast; in blocks 26.3, 26.10, 26.21, at 26.36 come from 13 block G, W^{-1}and from input devices, the output of these blocks

for j=k, is fed to the input of block 26.50 formation amount; output 26.11 block on 26.12 unit comes forecast value for k -1 point in time, in block 26.12 do the values of G, W^{-1}, y, output 26.12 block input 26.50 unit enters an expression analogous to (29), 26.22 block 26.25 block and 26.23 block receives the value is their prediction for k-2 time
in block 26.25 transmitted values G, W^{-1},, 26.25 block input 26.50 unit enters an expression analogous to (29), from the output of the block 26.23 comes forecast value for k-1 moment of time, in block 26.24 do the values of G, W^{-1}, y, output 26.24 block on 26.50 unit enters an expression analogous to (29), the inputs 26.37, 26.38, 26.39 blocks haveoutputs of these blocks have values for k-3, k-2, k-1 moment of time, which enter into blocks 26.41, 26.42, 26.40, respectively, in addition, the block 26.39 uses the results 26.38 block, and 26.38 block 26.37 block and, finally, 26.37 block are transferred to the initial values at the inputs of blocks 26.40,26.41, 26.42 do the values of G, W^{-1}, y, output blocks 26.40, 26.41, 26.42 block 26.50 do expressions analogous to (29); delay lines 26.4, 26.14, 14.28, 26.46 input blocks 26.5, 26.15, at 26.29, 26.47 transpose matrix functions and 26.6, 26.16, 26.30, 26.48 formation of the work is transferredafter which output blocks 26.6, 26.16, 26.30, 26.48 input blocks forming the difference of 26.7, at 26.17, 26.31, 26.49 enters the expressionat the output of which is formed the discrepancy of measurementsarriving at 26.51 shaping unit amounts; block 26.50 per unit 26.53 forming relationships is transmitted mn the value of
at the first sign 26.52 formation works denominator of the regularization parameter with unit 26.51 enters the expression

and to the second input unit 26.52 enters the value of Δt from the fifth information output unit 13 forming and outputting a priori data (figure 1), the output of block 26.52, we obtain the expression

From the first output unit 26.52 formation works denominator of the regularization parameter to the second input unit 26.53 form relationships comes the expression

The output of block 26.53 forming relationships, we get the value of the regularization coefficient α_{n+1}.

The first shaping unit works the numerator of the regularization parameter 26.3 (figure 6) works as follows. To the input unit 26.3.1 formation matrix function receives the value of the information processwith unit output 26.3.1 input blocks 26.3.2 the formation of the partial derivatives of matrix functions and block 26.3.4 comesand then , with unit output 26.3.2 values are sent to the input unit 26.3.3 transpose matrix functions, where the input unit 26.3.5 formation works comeson the other inputs are values of G^{T}, W^{-1}and C is Uchenie
from the output of the block 26.3.4, one input of which comes from the value G^{T}to the input of the block 26.3.5 output unit 26.3.6 transpose matrix functions, on which input respectively from the output 13 of the block receives the value of G; output block 26.3.5 have the value ofwhich is supplied respectively to the input blocks 26.3.7 transpose matrix functions and 26.3.8 formation works, the output will have a value

The first computing unit forecast is as follows. Input blocks 26.11.4 the formation of matrix functions and 26.11.3 formation of partial derivatives receives the value of the information processwith unit output 26.11.4supplied to the other input of the block 26.11.3 and blocks 26.11.5 formation works, 26.11.6 formation amount output unit 26.11.3to the input of the block 26.11.1 formation works, the other input of which receives the values of G_{n}(k) and Δt, the output of this unit to the input unit 26.11.2 formation amount transferred isto another input of the block 26.11.2 enters a value of G_{n}(k)in the output 26.11.2 block whose output is the first output block is ycycline forecast
have forecast G_{n}(j),j=k+1; the second input of block 26.11.5 enters the value of Δt, the output of which issupplied to the second input unit 26.11.6 whose output is the second output of the computing unit of the forecast, the output of which have x_{ηn}(j). Other units of the calculation of the forecast operate in the same manner.

Sources of information

1. Ago, Washkansky. Applied questions of optimal linear filtering. - M.: Energoizdat, 1982. - 96 S.

2. Aparina, Fluder. Digital processing of radar data. Support goals. - M.: Radio and communication, 1993. - 118 S.

3. Patent No. 2362255. Russia. 2009. Digital intellectual recursive filter. // Kostoglotov A.A., Kuznetsov A.A. Ponomarev, A.S., Lazarenko SV

4. Kostoglotov A.A. Synthesis of intelligent measurement procedures based on the principle of regularization A. N. Tihonov's // Measurement techniques, No. 1, 2001. P.8-12.

5. Kostoglotov A.A. Method of successive approximations in theory of filter // Automatic control and computer engineering, No. 3, 2000. P.53-63.

6. Kostoglotov A.A., Kuznetsov A.A. Synthesis of intelligent measuring procedure based on the method of minimum error // Measurement techniques, No. 7, 2005. P.8-13.

7. Tikhonov A.N., Arsenin VIA Methods for solving ill-posed problems. - M.: Nauka, 1986. - 288 S.

8. Vasiliev FP Methods for solving extreme ass is H. - M.: Nauka, 1981. - 106 S.

9. Kostoglotov A.A. Digital smart metering procedure // Measurement techniques, No. 7, 2002. P.16-21.

10. Detistov VA, Taran NR. Synthesis of optimal control of a gradient method based on a prediction of the model // And and So in 1990. No. 10. - P.46-56.

Digital intelligent multi-stage filter contains blocks: first, second, third, fourth blocks forming the difference; first, second, third, fourth correction blocks; block forming and outputting a priori data; first, second, third, fourth blocks forming amount; first, second, third, fourth, fifth, sixth, seventh and eighth blocks the formation of matrix functions; first, second, third and fourth delay lines; first, second and third block of the calculation of the regularization parameter, the first, second, third and fourth outputs of the processing unit and outputting a priori data are connected respectively with the second, third, fourth and fifth information inputs of the first, second, third and fourth correction block, in addition, the first, second, fifth outputs of the block forming and outputting a priori data are connected respectively with the third, second and fifth information inputs of the first, second and third block of the calculation of the regularization parameter, the first data output of the first correction block connected to the first information input of the first shaping unit amounts the output of which is connected to the information input of the first shaping unit matrix functions and fourth information input of the first block of the calculation of the regularization parameter, the first data input which is the input device, the output of the first block of the calculation of the regularization parameter is connected with the ninth informational input of the second error correction block, the output of the first processing unit matrix functions connected with the information input of the first delay line, the output of which is connected with the second information input of the first shaping unit amounts, with the seventh information input of the first error correction block and the information input of the second shaping unit matrix functions, the second information, the output of which is connected with the sixth information input of the first error correction block, the first information output of the second processing unit matrix functions connected with the second information input processing unit difference, the output of which is connected to the first information input of the first error correction block; a second information output of the first correction block is connected with the eighth informational input of the second error correction block, the first information output of which is connected to the first information input of the second processing unit amount, the output of which is connected to the information photostrictive block the formation of matrix functions and with the fourth information input of the second block of the calculation of the regularization parameter, the output of which is connected to the ninth informational input of the third error correction block, the output of the third processing unit matrix functions connected with the information input of the second delay line, the output of which is connected with the second information input of the second processing unit amount, the seventh information input of the second error correction block and the information input of the fourth processing unit matrix functions, the second information, the output of which is connected with the sixth information input of the second error correction block, the first data output of the fourth processing unit matrix functions connected with the second information input of the second processing unit difference, the output of which is connected to the first information input of the second error correction block; a second information output of the second block correction connected with the eighth informational input of the third error correction block, the first information output of which is connected to the first information input of the third processing unit amount, the output of which is connected to the information input of the fifth processing unit matrix functions, the output of which is connected to the information input of the third delay line, the output of which is connected with the second information input of the third unit forming the sum, the seventh information input of the third block adjusting the AI and information input of the sixth processing unit matrix functions, the second information output of which is connected with the sixth information input to the third error correction block, the first data output of the sixth processing unit matrix functions connected with the second information input of the third processing unit difference, the output of which is connected to the first information input of the third error correction block; a second information output of the third error correction block is connected with the eighth informational input of the fourth correction block, the first information output of which is connected to the first information input of the fourth processing unit amount, the output of which is an output device connected to the information input of the seventh processing unit matrix functions, the output of which is connected to the information input of the fourth delay line, the output of which is connected with the second information input of the fourth processing unit amount, the seventh information input of the fourth block of the correction and the information input of the eighth block the formation of matrix functions, the second information, the output of which is connected with the sixth information input of the fourth correction block, the first data output of the eighth block the formation of matrix functions connected with the second information input of the fourth processing unit difference, the output of which is connected to p the pout information input of the fourth correction block, in the device structure of the block of calculation of the regularization parameter is introduced: the first, second, third and fourth blocks forming the difference and the shaping unit works denominator of the regularization parameter that define the new connection, the first information input of the first processing unit difference is connected with the fourth information input of the first shaping unit works the numerator of the regularization parameter, and the second information input of the first processing unit difference is connected with the first information output of the first shaping unit works; the first data output of the first processing unit difference is connected with the fourth information input of the first shaping unit amount; the first information input of the second block forming the difference is connected with the fourth input information the first shaping unit works the numerator of the regularization parameter, and the second information input of the second block forming the difference is connected with the first information output of the second shaping unit works; the first data output of the second processing unit difference is connected with the third information input of the first shaping unit amount; the first information input of the third block forming the difference is connected with the fourth information input is m the fourth processing unit works the numerator of the regularization parameter, and the second information input of the third block forming the difference is connected with the first information output of the third processing unit works; the first data output of the third processing unit difference is connected with the first information input of the first shaping unit amount; the first information input of the fourth block forming the difference is connected with the fourth information input of the seventh processing unit works the numerator of the regularization parameter, and the second information input of the fourth block forming the difference is connected with the first information output of the fourth block formation works; the first data output of the fourth unit forming the difference is connected with the second information input of the first processing unit; a first information input processing unit works denominator of the regularization parameter is connected with the first information output of the first processing unit amount, and the second information input is connected to the fifth information output unit for forming and issuing a priori data, the first information output processing unit works denominator of the regularization parameter is connected with the second information input unit formation relations.

**Same patents:**

FIELD: physics.

SUBSTANCE: device has an adder, an inverter unit, two comparator units, a first AND element, a first bidrectional counter, a unit for setting deviation ratio, having a register, a counter and a delay element, two AND elements, a second bidirectional counter of the smoothing result, an inverter, a dynamic characteristic control unit, having a first and second pulse former, an OR element, a counter, a main and additional AND element and a flip-flop, a single-channel smoothing unit, having an adder connected in series to a register, the output of which is connected to the input of the second bidirectional counter of the smoothing result, the bus for recording into the register is connected to the clock input of the device, and the input of the adder is connected to the data input of the device.

EFFECT: high accuracy of smoothing in dynamic conditions.

3 dwg

FIELD: information technologies.

SUBSTANCE: invention relates to specialised computer means and may be used in systems, where hardware implementation of algorithms of digital signal filtration is required. For instance, when zero level is assessed at the background of pulse signals/noise or under conditions of asymmetrical limitation of dynamic range, relative to zero level. Digital filter comprises digital comparator (1), the first and second reversive counters (2, 3).

EFFECT: simplified hardware implementation of digital filter for assessment of slowly changing median of signal.

1 dwg, 1 tbl

FIELD: information technology.

SUBSTANCE: invention can be used to design a system-wide hardware environment of digital autoregressive filters and finite impulse response (FIR) filters, devices for identification, convolution and modular calculations. The device has a clock pulse generation unit, a filter type code storage unit, a filter polynominal structure code storage unit, a filter unit consisting of n series-connected identical elements - cells which form a homogeneous programmable filter register medium.

EFFECT: possibility of programmable tuning of a digital filter for carrying out functions of an autoregressive or FIR filter using an arbitrary filter polynomial.

3 dwg

FIELD: computer engineering.

SUBSTANCE: invention relates to digital computer engineering and can be used in digital signal processing systems for optimum nonlinear filtering. The device contains six units for generating matrix functions, three correctors, three units for generating difference, three units for generating sum, three delay lines, unit for generating and outputting prior data, and two units for calculating regularisation parametre.

EFFECT: increased accuracy of evaluating information process parametres in measuring systems.

8 dwg

FIELD: information technology.

SUBSTANCE: present invention relates to digital computer technology and can be used in systems for digital processing radio signals for optimum non-linear filtration. The device has blocks for generating matrix functions (4, 6, 10, 12), corrector units (2, 8), differential generating units (1, 7), summing units (3, 9), delay line (5, 11), unit for generating and output of priori data (13). The device also has a unit for calculating regularisation parametre (14), which is linked to the rest of the units.

EFFECT: more accurate evaluation of the information process in measuring systems.

8 dwg

FIELD: computer science.

SUBSTANCE: device has sum forming blocks, matrix functions forming block, difference forming block, delay lines, apriori data output block.

EFFECT: higher precision.

6 dwg

FIELD: electricity.

SUBSTANCE: electric element having ports and linear electric properties characterised in matrix, which is impedance matrix, admittance matrix or dissipation matrix of electric element and connecting voltage applied to ports with current passing through these ports. Electric element has inactivity determined by means of parameters perturbation up to perturbated set of parameters provided that this perturbated set of parameters corresponds to function in Boolean values.

EFFECT: increase of forecast accuracy for technically relevant linear electric properties of electric elements.

16 cl

FIELD: information technology.

SUBSTANCE: digital predicting and differentiating device includes a unit for estimating first derivatives, having a subunit for calculating a first derivative at a second (n-1)-th reference point of process history consisting of three adders, the output of which is the third data output of the device, and a subunit for calculating a first derivative at the third (n-2)-th reference point of the process history consisting of three adders and a block of inverters, the output of the subunit being the fourth data output of the device.

EFFECT: broader functional capabilities of prediction devices by obtaining estimations of first derivatives using numerical differentiation formulae for prehistory nodes of an input smoothed discrete sequence.

5 dwg, 2 tbl

FIELD: information technology.

SUBSTANCE: device has an image storage unit, a pixel storage unit, a turning unit, a vocabulary forming unit, a vocabulary storage unit, a processing unit, a priority calculating unit, turning unit, an adaptive form determination unit, a similarity search unit, a pixel averaging unit, an image filling unit and a clock pulse generator.

EFFECT: reconstruction of image pixel values under incomplete a priori information conditions.

6 dwg

FIELD: information technology.

SUBSTANCE: method of detecting and eliminating pulse noise when processing images involves comparing values of the original digital image with different threshold values. A set of penalties is then created for values of the original digital images exceeding the threshold values. The resultant penalty values are formed by adding separate penalty values for each reading. Readings whose resultant penalty values exceed the calculated threshold level are considered abnormal. Further, two-dimensional arrays of penalties are formed for each threshold level. Two-dimensional nonstationarity regions are determined and localised by a two-dimensional programmed detector with subsequent zeroing of the penalty values. The detected pulse noise values are eliminated by replacing them values of a first-order approximating surface on the localised regions.

EFFECT: detection and elimination of pulse noise values when processing digital images in conditions of non-parametric expected uncertainty of statistical characteristics of pulse noise and the image.

2 cl, 4 dwg

FIELD: information technology.

SUBSTANCE: device has a smoothing unit consisting of an adder, inverters, comparators, counters, AND logic elements, a deviation ratio setting unit and a dynamic characteristic control unit, and a prediction unit comprising three subtractors, two prediction subunits and a register.

EFFECT: high accuracy of prediction and simplification of the device.

7 dwg, 2 tbl

FIELD: information technology.

SUBSTANCE: apparatus for processing two-dimensional signal when reconstructing images has an image storage unit, a pixel storage unit, a directory creating unit, a directory storage unit, a processing unit, a priority calculation unit, an adaptive form determining unit, a resemblance search unit, a pixel averaging unit, an image filling unit and a clock-pulse generator.

EFFECT: reconstruction of image pixel values with incomplete prior information.

5 dwg

FIELD: physics.

SUBSTANCE: function codes are rounded-off to the nearest level and the obtained codes are stored. The optimality criterion code is calculated and stored. Starting with a certain initial number L of the function code, the direction of rounding-off this code is changed and the optimality criterion code is calculated. If the optimality criterion code falls, the changed value of the code is stored and a new value of the optimality criterion code is calculated and stored, otherwise the initial L-th function code and the initial optimality criterion code are stored, and calculation is moved on to the next number of the function code L+1, where it is checked whether the optimality criterion code falls in the same way as that when the L-th function code was changed. Further, the process is continued until the optimality criterion code does not fall in a sequence of n function codes, read from the code value in which the last fall in the optimality criterion code took place.

EFFECT: reduced absolute error in the amplitude of the reproduced sinusoidal signal.

FIELD: information technology.

SUBSTANCE: device has a unit for storing input realisation, switches, approximation units, estimation storage units, arithmetic adder, a unit for storing useful component estimates, a control unit, a delay unit, a clock-pulse generator, two units for breaking down into intervals, each having a random number generator, a unit for averaging related values, a ranging unit and a register for storing random number samples. The control unit has a shift register for sampling column random numbers, a shift register for sampling row random numbers, a delay unit for sampling column random numbers, a delay unit for sampling row random numbers, a counter and a unit for checking conditions.

EFFECT: two dimensional estimation of the useful component in conditions with insufficient prior information on statistical characteristics of additive noise and useful component function.

2 dwg

FIELD: information technologies.

SUBSTANCE: device comprises serially connected frequency filter, digitiser and unit of reduction to perfect instrument (RPI), intended for interpolation of counts supplied to its inlet, detection of weight of basic final duration of signals in inlet signal on the basis of interpolated counts decomposition into Fourier series by orthogonalised reactions of frequency filter into basic signals and for formation of outlet signal as a superposition of basic signals with account of their weight in inlet signal, besides versions of device include connection of noise suppression unit or serialy connected unit of signal growth speed assessment and normalisation unit between digitiser and RPI unit.

EFFECT: improved resolution and sensitivity to elements of signal, increased efficiency and simplification of device for signals processing.

4 cl, 12 dwg

FIELD: information technologies.

SUBSTANCE: device comprises unit of input realization storage, clock oscillator, control unit, unit of useful signal extraction, unit of storage of five last values of useful component assessment, unit of approximation with polynom of the first degree, unit of approximation with polynom of the second degree, unit of output realization storage. In device end values of assessment are approximated with the help of method of least squares with polynom of the first or second degree, then produced equation of assessment is used to calculate values in forecast points.

EFFECT: forecasting measurement results on the basis of useful signal extraction without end effects, under conditions of limited a priori information about useful and accidental component.

1 dwg

FIELD: computer science.

SUBSTANCE: device has sum forming blocks, matrix functions forming block, difference forming block, delay lines, apriori data output block.

EFFECT: higher precision.

6 dwg