Method for adaptive pid law-based control and system for realising said method

FIELD: physics; control.

SUBSTANCE: invention relates to automated control of complex information devices using PID control laws, and can be used in radio systems with chaotic dynamic realisation of their target functions in intense information perturbation conditions. The method involves establishing conformity between acceptable criticality levels of deviations of configuration parameters of the controlled system and the criticality level of their deviations; using values of a decision matrix to calculate the error of deviations of values of the configuration parameters of the controlled system and checking their conformity with the acceptable criticality levels of deviations of the configuration parameters of the controlled system; in case of conformity, storing previous values of the configuration parameters of the controlled system and re-factoring the initial content of a plurality of values of the decision matrix, and in case of non-conformity, storing that event and calculating the error of deviations of values of the configuration parameters of the controlled system and assigning a control action for the controlled system based on selection of values from the decision matrix; in case of detection of missing and/or incorrect given values of configuration parameters of the controlled system, re-factoring content of a plurality of values of the decision matrix on each of the configuration parameters of the controlled system for the given operating conditions of the controlled system by adding the detected missing values and/or changes in incorrect values of the decision matrix; storing previous values of the configuration parameters of the controlled system, assigning a control action to the controlled system and re-factoring content of a plurality of values of the decision matrix by replacing previous values of the configuration parameters of the controlled system with current values of the configuration parameters of the controlled system. The system comprises switching matrices of inputs and outputs (1, 2) of the decision device (3), a proportional unit (4), functional logic controller (9), an actuating unit (11), a controlled object (12) and an integrating-differentiating unit (10), having K integrating-differentiating clusters (7₁…7_k), each having one of the K integrating units (5₁…5_k) and one of the K differentiating units (6₁…6_k) with own normalising factors.

EFFECT: improved dynamics of control processes and broader functional capabilities.

2 cl, 7 dwg

The claimed invention relates to the field of automated control of complex information devices using proportional-integral-differential laws (PID-laws) regulation, and can be used in radio systems with chaotic dynamics of the realization of their objective functions in conditions of intensive information indignation.

At present considerable practical interest is the construction of automatic control systems in the aspect of implementation ensure reliable operation of modern information and communication systems (ICS). This is primarily due to the fact that modern information and communication complexes has evolved into a complex information system, in most cases, chaotic realization of their objective functions. In this case the operation in the shared environment of acceptance, have a significant impact on the effectiveness of potentially dangerous code or analog sequences (in digital or verbal form) and regular mode of conflict of functioning as separate elements, and all IP require improvement of the management functions as the control algorithms included in the groups of objects, and control algorithms of the physical parameters of IP,such as current, voltage, frequency and other

Contents methods of adaptive control is that the system controller in any way monitors and diagnoses as the internal state of the control object, and the external disturbance input to the object, and fits the regulatory process for changing the state of the object. It should be noted that it is necessary to know the initial parameters of the control object, characteristic for each of the control objects. If the object does not change, then all the parameters of the PID controller remains unchanged. If the parameters of the object has changed, then the adaptive controller for several iterations finds new optimal parameters and replaces your old settings to the new ones. This ensures the constancy of the optimal settings of the PID controller for changing the parameters of the object of regulation. System operating according to this method of control is called self-tuning adaptive PID controller. For systems with chaotic dynamics, with increased requirements to the quality of regulation is necessary to develop systems with individual structure, based on the physics of the processes in the control object.

The known method of proportional control based on the change of the control signal depending on the different the t between the true signal and the set value, described in [1]. The General formula to calculate the value of the output signal of way proportional control following [1, s]:

$$\text{y}=\text{G*e}\text{, (1)}$$

where G is the gain;

e - the magnitude of the error (the difference between the reference and the true value) in the system of proportional control.

The disadvantage of proportional control is that when you change the control signal depending on the difference between the reference and measured values there is no mechanism to adjust to the suddenly changing external conditions.

The known method of proportional-integral-differential control (PID control), the system adds additional inputs carrying information about the previous state of the system described in [1, p.148]. The General formula to calculate the value of the output signal of the way the PID control is as follows:

$$y=G\left(e+I{\displaystyle \int edt+D\frac{de}{dt}}\right),\left(2\right)$$

where G is the gain;

e - error control, i.e. the difference between the desired and actual values;

I - the added value of the integral of the error;

D - add the value of the derivative of the error.

From way proportional control method PID control differs additional calculations of the integral and derivative (difference between States in adjacent time intervals). Since these parameters depend on time, adding operations of differentiation and integration in the control algorithm can improve some parameters. The proportional part (G*e) causes the output signal to follow the input (set preset value). The differential part of$$\left(D\frac{de}{dt}\right)$$reflects the rate of change of the error control and provides a response output signal to the rapid changes of the input to compensate for the influence of the "load". Integration (I∫edt) reflects the sum of errors for a certain period and compensates for slowly changing error.

The way the PID control requires some tweaking of the system, i.e. the choice of parameters (coefficients) of all three quantities. In the first stage, the adjustment procedure is istemi PID control is to set the gain (G) large enough to ensure high speed operation of the system. Then the value of the derivative (D) is large enough to reduce possible excessive amplification and oscillation. Finally, the integral gain (I) is also great for eliminating the steady state error. Most often, the coefficients are chosen experimentally.

The method of operation of a system that implements the control method according to the formula (2) described in [1, s], called the method of Ziegler/Nichols. The algorithm of the method consists of the following steps:

1. First disable circuit integrating and differentiating signals, which makes the control system in a system with proportional control.

2. Increase the gain up until the output are insignificant or ringing. This gain level is denoted by ().

3. Measure the period of oscillation (P).

4. This is followed by checking for the installation of the coefficients of the proportional gain (G), and amplification of the signals of the integral and derivative in accordance with the experimental ratios. If the control system is only proportional, then set G=0.5 K. If the system proportional-integrating, we set G=0.45, I=1.2/R. If the system PID control, ustanavli the ut G=0.6, I=2/R, D=R/8.

Obviously, this method of tuning systems with PID regulation requires additional correction for optimization. The resulting disadvantages are listed below:

- the difficulties associated with the measurement of the oscillation period;

- the problem of saturation is possible to calculate the signal at the output, which is a real Electromechanical system will never be reached;

- the presence of uncertainty management due to the time delay that occurs from the feeder control signal to the response of the sensor;

- the presence of sharp changes in the signal that are not amenable to regulation.

The closest to the technical nature of the proposed method is a way to control the generalized discrete PID controller described in [2, pp.261-272], which depending on the input parameter set to obtain a certain structure of the digital control adopted for the prototype.

Prototype method consists in the following steps.

The input variable u(kt) discrete regulator is described by the equation, which is convenient to think of three components - loop prediction for the reference value u_F1, the feedback circuit on the output of the process u_FBand circuit lead on the measured perturbation u_F2:

The perturbation w(kt) affects the process in accordance with the transfer function, dependent on the characteristic features of the controlled process. Operator feedback$$\frac{S\left(q\right)}{R\left(q\right)}$$includes the dynamics of the sensor output feedback process, and the operator proactive management$$\frac{V\left(q\right)}{}R\left(q\right)$$includes the dynamics of the sensor lead on the measured disturbance.

To calculate the controller parameters it is most convenient to represent the process in the form:

$$y\left(kt\right)=\frac{TB}{AR+BS}\times u_c\left(kt\right)+\frac{AR}{AR+BS}\times w\left(kt\right).\left(4\right)$$

The first member defines a gear operator from the reference value u_C(kt) to the output of the process y(kt) through the paths lead and feedback, and the second - gear operator from the disturbance w(kt) to the output y(kt) through the feedback loop.

The parameters of the polynomials a and b depend on the type of process and therefore are assumed to be constant, and the parameters of the polynomials R, S, T can be configured. The conversion controller parameters in the coefficients of the polynomials can be represented as follows:

$$\{\begin{array}{c}R\left(q\right)=q^2-\left(1+\beta \right)q+\beta \\ T\left(q\right)=K\left(1+\alpha \right)q^2-K\left(1+\beta +\alpha \beta \right)q+K\beta ,\left(5\right)\\ S\left(q\right)=K\left(1+\alpha +\gamma \right)q^2-K\left(1+\beta +\alpha \beta +2\gamma \right)q+K\left(\beta +\gamma \right)\end{array}$$

$$\alpha =\frac{t}{T_i},\beta =\frac{\text{T}}{T_d+tN},\gamma =\frac{{\text{T}}_{\text{d}}}{\text{t}}\left(1-\beta \right),\left(6\right)$$

where N is the normalizing factor for a constant time;

K - parameter of the regulator gain;

T_i- the time constant of integration;

T_d- time constant differentiation;

q - operator save the previous value of the signal.

The algorithm of functioning of the discrete generalized controller 15 described in [2, pp.272], is shown in figure 1 in the form of a flowchart. When the unit 15.1 is the initialization state vectors and the calculation of the constant coefficients of the controller. At the time (kt) in block 15.2 read the values of the signals u_c(kt)w(kt), y(kt). Then in block 15.3 is the calculation of the control signal u(kt) based on the values of previous samples according to the formula:

$$u\left(kt\right)=T{u}_c\left(kt\right)-Sy\left(kt\right)+\left(1-R\right)w\left(kt\right)+x\left[\left(k-1\right)t\right],\left(7\right)$$

where T, S, R are calculated based on the constant coefficients of the controller parameters.

Update the state vectors u_c(kt), u(kt)y(kt) and the issuance of the values u(kt) is in the block 15.4.

As soon as the computed control signal, in block 15.5 is the calculation of the new sample values for use in the next step.

However, in the method prototype can be noted the following disadvantages:

- when switching modes of operation parameters of the control signal and the regulator must be installed manually;

because all transfer functions are determined by the characteristics of the process, pre-emptive signal is completely determined by the system model. If the model is inaccurate, pre-emptive signal may not completely offset the effect of the disturbance.

is not taken into account the characteristics of information systems.

The purpose of the system that implements the control method according to the law of the PID is in control of some physical quantity (e.g. voltage, frequency, or temperature of an object). Unlike the most common analog systems, digital control system may play a more complex function, and to generate a signal, which is not only a function of the input and output, but also the previous States of the input and output coefficients temporary changes, load conditions, etc. the Practical implementation of this control method requires the use of modern is elektronnoy base, including microprocessors, logic devices, digital interfaces.

A device for automatic regulation of "Supervisory proportional-integral-differential controller"described in [3], which allows simultaneous high-quality work and setting and disturbance and provides a significant improvement of the dynamic characteristics of the systems of control of technological parameters during processing of perturbations on different channels.

The disadvantage of the [3] is that dynamic blocks, providing a Supervisory part of the regulator, are constant values and are selected in advance to compensate for the disturbance, depending on the controlled parameter that limits the scope of such controls.

Known automatic control device objects with unknown parameters and unknown but bounded external disturbance "Self-tuning PID controller described in [4], providing the system with larger changes in the parameters of the object when switching from mode to mode.

The disadvantage of the [4] is the lack of adaptability to non-linear processes in the system with chaotic dynamics, due to the implementation of the device on the industrial controller. In this is case, configure the PID controller is given in advance formulas for certain characteristics of the object and is not optimal.

Classical automatic control system that implements the control method according to the formula (2)described in [1, s-150]. In the system PID control added an additional input that carries the information about the previous state of the system. The difference between the reference and measured values increases. Derivative and integral enhanced difference is summed with the amplified error signal, producing a signal at the output.

However, in the system [1] it can be noted the following disadvantages:

- requires additional correction of the adjustment parameters for optimization of work;

- the lack of regulation of abrupt changes of the signal.

The system adopted for the prototype and implementing the control method according to formulas (3)to(7), described in [2, s].

Structural-functional scheme of the system is shown in figure 3, where the following notation:

4 - proportional block (PB);

5 - integrating unit (IB);

6 is a differentiating block (DB);

11 - Executive block (ISB);

12 is a managed object (UO);

13 - the first adder (C1);

14 - the second adder (C2).

Functional diagram of the integrating unit 5 are presented in figure 4, where the following notation:

5.1 - integrating controller (HP);

5.2 sensor feedback (DOS).

Functional diagram of the differentiating unit 6 are presented on figure 5, g is e, the following notation:

6.1 - derivative controller (PD);

6.2 sensor perturbations (DV).

The system prototype contains a proportional unit 4 performing ahead of the reference value, the output of which is connected to the first input of the first adder 13; an integrating unit 5, a feedback, the output of which is connected to the third input of the first adder 13; a differentiating unit 6, exercising the pre-emption of a perturbation, the output of which is connected with the second input of the first adder 13, the output of which is connected to the input of the control unit 11, realising a total correction input velichanii its output connected to the input of the control object 12, the output of which is connected to the first input of the second adder 14, the output of which is connected to the input of the integrating unit 5.

When included in blocks 4, 5, 6 is to set the values of the constant coefficients of the controller. In the next moment in block 4 is fixed signal value u_c(kt), in block 5 is fixed to the value of the signal y(kt), in block 6 and the adder 14 is fixed the value of the signal w(kt). The state of the recorded signals is remembered for the next step of the calculation at the next time interval. Then transformed based on the constant coefficients of the parameters of the control signals fed to the input of the first adder 13, in which Vice which is the control signal u(kt) based on the values of previous samples according to the formula (7). Further, the output of the first adder 13 the value of the signal u(kt) is fed to the input unit 11 with further converted in accordance with the system requirements of the signal unit 12, the output of which the signal arrives at the input unit 14. Then, the output unit 14, the signal received at the input unit 5, which provides compensation for slowly changing bias control.

Further steps are repeated, starting with step of fixing the values of the signals u_c(kt)w(kt), y(kt).

However, in the system prototype can be noted the following disadvantages:

- when switching modes of operation parameters of the control signal and the regulator must be installed manually;

because all transfer functions are determined by the characteristics of the process, pre-emptive signal is completely determined by the system model. If you use a less accurate model, pre-emptive signal may not completely offset the effect of the disturbance.

A common disadvantage of the above systems that implement the control of the PID act, are poor indicators of quality in the control of nonlinear and complex systems, and in the absence of sufficient information about the control object.

The task, which directed the claimed invention is to improve the performance of regulators in terms of working with significant Uro is it a priori uncertainty of the parameters with the purpose of providing additional protection in terms of special effects and intentional interference.

Technical result achieved - improved dynamics of the processes of regulation and extension of functional capabilities, which are as follows:

- ensuring high quality and accuracy control of the controller, including the measurement of very small quantities or values in a very wide dynamic range;

- providing opportunities for repeated adjustment of control parameters;

- providing opportunities for rapid correction of sets of constant values of the matrix of decision making, which increases the overall intelligence of the dynamic system control;

- improvement of the dynamics of regulatory processes due to the implementation of the sequential (cascade) connection cluster units;

- increased reliability in reducing the influence of factors not considered ICT environment on the stability of open dynamic systems with a pronounced oscillatory dynamics or parameters that are rapidly changing in time;

- allows selection of the desired dynamic branching processes and create a rational structure by controlling the initial data and the choice of the ensemble of States most appropriate allowed by the system.

To solve the problem stated how adaptive is about control on the basis of the law of the PID, consisting in the following steps:

using a priori parameter values configuration of a managed system (MOUSTACHE) for the specified operating conditions CONDITION, determine acceptable levels of criticality deviations of the relevant configuration settings for the CA in the form of logical functions defined on respective sets of deviations of the values of the configuration parameters of the CA, according to the possible levels of the deviations of the input signal from the corresponding values of the matrix of decision making;

the obtained values of acceptable levels of criticality deviations configuration CONDITION is used to install one correspondence between the parameter configuration CONDITION and level of criticality reject;

the values of the matrix of decision making compute the error variance of the values of the configuration parameters MUSTACHE;

if the calculated error variance values of the configuration parameters CONDITION correspond to a valid severity of deviations configuration settings MUSTACHE, and retain the previous settings configuration MOUSTACHE and spend refactoring of the initial content of the set of values of a matrix of decision-making;

if the calculated error variance values of the configuration parameters CONDITION does not correspond to a valid severity of the deviation, parametro the configuration of the CA, remember this event and the calculated error variance values of the configuration parameters MUSTACHE;

next, perform the adaptation values of the configuration parameters of the CA to the identified deviations functioning CONDITION by assigning a controlling influence on US on the basis of selecting values from a matrix of decision making, when it set up a correspondence between current and valid configuration values US;

if the identified missing and/or incorrect settings that you have defined the configuration of the CA, perform the refactoring of the content of the set of values of a matrix of decision making for each of the configuration parameters CONDITION for the specified operating conditions CONDITION by adding the identified missing values and/or changes to incorrect values of the matrix of decision making;

then retain the previous settings configuration MUSTACHE, appointed to manage the impact on US and spend refactoring content of the set of values of a matrix of decision making by replacing the previous values of the configuration parameters of the CA current values of configuration parameters MUSTACHE;

repeat the above steps from step definitions severity deviations configuration CONDITION for the specified operating conditions MUSTACHE.

To solve this problem for what is the adaptive control system according to the PID act, contains a switching matrix inputs (Kmwh) and outputs (Cmwyj), connected in series solver, the Executive unit and the managed object whose output, which is the output device, connected to the second signal input Kmwh, the first signal input which is the input of the controlled signal; a controller functional logic (CFL), is proportional to the unit and integramouse-differentiating unit (IDB), including To integramouse-differentiating clusters (IDK), each of which contains at integrating and differentiating unit with its own normalizing coefficients, and pairs of signal inputs IDB are pairs of signal inputs corresponding IDK where the first signal input IDK signal is input corresponding to the integrating unit and the second signal input IDK is the signal input of the corresponding differentiating unit; output pairs IDB are pairs of outputs corresponding IDK, from which the first output is the corresponding output of the integrating unit, and the second output is the corresponding output of the differentiating unit; a first output Cmvh connected with the signal input is proportional to the unit, the output of which is connected to the first signal input Cmwyj, the first output of which is connected with the first signal input re the surrounding devices, the remaining pairs of the signal inputs of which are connected with corresponding To pairs of outputs Cmwyj; the remaining pairs of outputs Cmvh connected to respective pairs To the signal inputs IDB TO pairs of outputs which are connected with the respective pairs To the signal inputs of Cmwyj, group outputs, which by means of the signal bus is connected with a group of signal inputs Cmvh; the first output of CFLs is connected with the control input Kmwh, the second output with the control input is proportional to the block, the third output - control inputs To each of the integrating and differentiating blocks; the fourth output with the control input Cmwyj, and the remaining outputs are connected CFLs with the respective control inputs of a casting device, the information outputs of which are connected to the appropriate information inputs CFLs.

Before describing the implementation of the claimed inventions, it is necessary to give the following explanation.

The term "cluster analysis" to denote a set of methods, approaches and procedures that are designed to solve problems of formation of a homogeneous component in an arbitrary problem domain and intended to find a partitioning of the researched population into subsets are relatively similar, similar objects [5]. The need for analysis and use of large volume of the MOU objective and subjective information, associated with the informal and poorly formalized tasks of different physical nature, required extensive development and implementation of new ways of applying these methods. Cluster analysis methods do not use any a priori assumptions about the probabilistic nature of the initial information and shall be guided only by heuristic considerations on the nature and characteristics of the study population. The facility attributable to the same cluster should be more similar to each other than objects from different clusters.

For control systems of complex information objects required to define more and information structure of the system, i.e. the number and location of additional channels of communication object control device, since such systems are usually built as multi-circuit and compensation of disturbances.

The solution of the problem of adaptive control is carried out by performing the following actions.

Pre-populated matrix of decision making in accordance with the required parameters MUSTACHE. To do this, select the ranges of input and output signals, the shape of the membership function of the desired parameters, rules, and inference mechanism, the ranges of the scale factors.

Then set the values for the initial approximations of coefficient is To, T_i, T_d. in accordance with formulas (5), (6). This can be done either by Ziegler/Nichols [1, s], or based on the requirements of specific tasks.

Next, formulate a criterion function that is required for finding the optimal values of the tuning parameters optimization methods. The optimization criterion function can be performed using genetic algorithms, methods of theory of fuzzy sets or the use of the principles of neural network training [13-18]. Optimization of fuzzy settings can be performed, for example, in the simulation result management process specified parameters [13-18].

After optimization determine the restrictions on the range of changes of configuration parameters of membership functions. As a membership function can be used piecewise-defined functions, describing isosceles triangles.

The obtained data store in a table of constants in the matrix of decision making. At this stage the preliminary determination of the parameters of the system ends.

Using a priori parameter values configuration of a managed system (MOUSTACHE) for the specified operating conditions CONDITION, determine acceptable levels of criticality deviations of the relevant configuration settings for the CA in the form of logical functions defined on respective sets the variance of the values of the configuration parameters MOUSTACHE, according to the possible levels of the deviations of the input signal from the corresponding values (constants)that are stored in the matrix of decision making.

The obtained values of acceptable levels of criticality deviations configuration CONDITION is used to install one correspondence between the parameter configuration CONDITION and level of criticality reject it.

Then the values of the matrix of decision making compute the error variance of the values of the configuration parameters MUSTACHE.

If the calculated error variance values of the configuration parameters CONDITION correspond to a valid severity of deviations configuration settings for US, this means not detected deviation parameters configuration and adaptation of the managed system is not required. In this case retain previous settings configuration MUSTACHE (fixed configuration CONDITION without changing the control action).

Then retain the previous settings of the system and carry out the refactoring of the initial content of the set of values of a matrix of decision making based on long-term analysis (called refactoring technique to improve the structure or sequence of the data without changing the functionality by adding, replacing or deleting objects [6]).

If the calculated error discard the deposits of configuration values US not correspond to a valid severity of deviations configuration settings MOUSTACHE, this means that the observed deviation from the specified operating conditions and the required adaptation CONDITION. In this case, remember that the deviation of the calculated error variance values of the configuration parameters MUSTACHE.

Next, perform the adaptation values of the configuration parameters of the CA to the identified deviations functioning CONDITION by assigning a controlling influence on US on the basis of selecting values from a matrix of decision making. This establishes the correspondence between current and valid configuration values MUSTACHE, that is, for each of the parameters are mapped to the current values and the values of the possible number of valid parameter values MUSTACHE.

If the identified missing and/or incorrect settings that you have defined the configuration of the CA, perform the refactoring of the content of the set of values of a matrix of decision making for each of the configuration parameters CONDITION for the specified operating conditions CONDITION by adding previously identified missing values and/or changes to incorrect values of the matrix of decision making.

Then retain the previous settings configuration MUSTACHE, appointed to manage the impact on US and spend refactoring content of the set of values of a matrix of decision making based on long-term analysis, in accordance with Uslon the regulations of a given algorithm by replacing the previous values of the configuration parameters of the CA current values of configuration parameters MUSTACHE.

Then repeat the above steps from step definitions severity of deviations configuration settings MUSTACHE from a given value for the given operating conditions MUSTACHE.

The inventive method is illustrated using figure 2, which presents the algorithm of adaptive control by the PID act, implemented by the functional logic controller (CFLs) 9 and supported by the architecture of the inventive system, the structural-functional scheme is shown in Fig.6, where the following notation:

1 - switching matrix inputs (Kmwh);

2 - switching matrix outputs (Cmwyj);

3 - critical device (RU);

4 - proportional block (PB);

5₁...5K - integrating blocks (IB);

6₁...6K - differentiating blocks (DB);

7₁...7K - integramouse-differentiating clusters (IDK);

8 - signal bus (school);

9 is a functional logic controller (CFLs);

10 - integramouse-differentiating unit (IDB);

11 - Executive block (ISB);

12 is a managed object (UO).

The inventive system includes a switch matrix inputs (Kmwh) 1, the first output of which is connected to the signal input of the proportional block (PB) 4, the output of which is connected to the first signal input switch matrix outputs (Cmwyj) 2, the first output of which is connected to the first signal input of the decision of the feeder (RU) 3.

Integramouse-differentiating unit (IDB) 10 of the claimed system provides To integramouse-differentiating clusters (IDK) 7₁...7K, each of which contains one To integrating units 5₁...5K with its own normalizing factor And₁...IR and one For differentiating blocks 6₁...6K with its own normalizing coefficient D₁...DK, respectively.

And in IDB 10 has To pairs of signal inputs (i.e. 2K signal inputs), which at the same time are pairs of signal inputs corresponding IDK 7₁...7K, where the first signal input IDK 7 is a signal input of the corresponding IB 5, and the second signal input IDK 7 is a signal input of the corresponding 6 DB. In addition, IDB 10 has To pairs of outputs (i.e. 2K outputs), which at the same time are pairs of outputs corresponding IDK 7₁...7K, where the first output IDK 7 is the output of the corresponding IB 5, and the second output IDK 7 is the output of the corresponding 6 DB.

When the pairs of signal inputs IDB 10 are connected respectively To pairs of outputs Cmvh 1, and To output pairs IDB 10 connected to respective pairs To the signal inputs of Cmwyj 2, To pairs of outputs which are connected with the respective pairs To the signal inputs of a casting device (PY) 3, the output of which through the Executive unit (IB) 11 is connected to the input of the controlled object (UO) 12, the output of which being the output device, connected to the second signal input Cmvh 1, the first input of which is adjustable input signal.

Also, the inventive system includes a controller functional logic (CFL) 9, the first output of which is connected with the control input Cmvh 1, the second output with the control input PB 4, the third output - control inputs To each of the integrating 5₁...5K and differentiating 6₁...6K blocks; the fourth output with the control input Cmwyj 2, and the remaining outputs CFLs 9 connected to respective control inputs RU 3, the information outputs of which are connected to the appropriate information inputs CFLs 9.

In addition, the group outputs Cmwyj 2 by means of the signal bus (school) 8 is connected with a group of signal inputs Cmvh 1.

In the General case, the PID control is a nonlinear transformation:

$$u_{\sum }={\displaystyle \underset{i}{\sum }{\lambda }_i{u}_i}+{\displaystyle \underset{i,j}{\sum }{\lambda }_{ij}{u}_i\left(u_j\right),\left(8\right)}$$

where λ_i,λ_ij- the weight coefficients of the input variables u_iu_j;

u_i=u_i(kt,Λ), u_i(u_j)=u_i(kt,u_j(kt,Λ)) - vectors of values of the control signal, which is defined by the formula (7);

Λ=f(u_F1u_FBu_F2- function coefficients, calculated on the basis of the parameters of the controller.

As mentioned above, the algorithm of adaptive control by the PID act is implemented by the block CFLs 9 (see figure 2).

When you switch the unit 9.1 initializes the values of the coefficients of regulators And₁...IR and D₁...DK by feeding the control signals on the control inputs of ID 10, and performs the setting PB 4 and initialization patterns Cmvh 1, Cmwyj 2 by submitting to their control inputs control signals.

Next, in block 9.2 reads the values of the input signals of the RU 3, then for each configuration parameter CONDITION occurs, the appropriate level of criticality reject it in the form of a logic function defined on the set of a number of variance parameters configuration CONDITION according to possible deviation levels of the input signal from the target value stored in the table of constants in the matrix of decision making.

For the eat in box 9.3 determine the suitability of the variance parameters configuration CONDITION for constant matrices a decision.

In block 9.4 checks to identify deviations configuration settings MUSTACHE and identify necessary corrective actions on US.

In case of negative result of the check, the system switches to the unit 9.5, where fixed configuration CONDITION without changing the control then passes to block 9.9.

In case of a positive test result, the system switches to the unit 9.6, where the state, the rejection and assign the control action based on the selection of the constants of the matrix of decision making to adapt US to the identified deviations operation, for each of the parameters are mapped to the current values and the values of the possible number of valid parameter values MUSTACHE, then move to block 9.7.

In block 9.7 checks to identify missing and/or incorrect set of elements from a variety of a number of configuration settings MUSTACHE.

If you identify missing and/or incorrect set of elements, a transition is made to block 9.8, which is refactoring (restructuring) initial content set for each of the disturbed conditions in accordance with known configuration requirements by adding identified in the previous step, missing values and/or changes correctly.

If not detected missing/what if incorrectly specified elements, moves to block 9.9, where retain the previous settings MUSTACHE.

Then in block 9.10 calculate the control signals for all control loops, identify and generate new values for RU 3. Specific factors calculations are determined by the nature of the adjustable MUSTACHE.

Next, in block 9.11 spend refactoring initial content set based on the long-term analysis in accordance with the conventional rules of a given algorithm, by replacing the previous values of the configuration parameters of the CA current values of configuration parameters MUSTACHE.

Then repeat the above steps, starting with the block 9.2.

The system that implements the method of adaptive control by the PID act, works as follows.

When you enable CFLs 9 initializes the values of the coefficients And₁...IR and D₁...DC blocks 5 and 6, produces the setting PB 4 and initialization patterns of connections Cmvh 1, Cmwyj 2 by feeding on the control inputs of these blocks corresponding control signals.

Next CFLs 9 reads the initial values of all signals from the information output PY 3. The data obtained keep in CFLs 9 for the subsequent calculations.

Input adjustable signal u_c(kt) and output signal y(kt) are received respectively in the first and second signal input is Cmvh 1. In addition, in real conditions, the system is under the influence of signal perturbation w(kt).

Blocks Cmvh 1, Cmwyj 2 organize necessary for the implementation of the target function system connection IDK 7₁...7K into a single structure, which provides the input signal u_c(kt) signal inputs IB 4 and IDB 10.

In PB 4 is proportional to the conversion of the input signal pre-emption for the reference value) in accordance with previously established parameters; then the converted output signal PB 4 through Cmwyj 2 is fed to the first signal input ROUX 3.

Block 10 is integral-differential conversion of the input signal, consisting in the implementation of feedback and proactively perturbation. The converted signal outputs IDB 10 through Cmwyj 2 is supplied to the other signal inputs RU 3.

School 8, forming the interface between Cmwyj 2 and Kmwh 1 allows you to organize serial IDK 7₁...7K (connects outputs IDK 7₁...7K to the inputs IDK 7₁...7K if necessary, the implementation of sequential patterns). For the implementation of consistent patterns across school 8, you must use the (2K-1) pairs of connections.

Thus the number of signal inputs and outputs of blocks 1, 2, 3, and 10 is determined by the task from the management and the degree of uncertainty of the parameters of the controlled object.

Each IB 5₁...5K and 6 DB₁...6K has its own normalizing coefficient, calculated in CFLs 9 on the basis of previous signal values. Proactive management is implemented in 6 DB₁...6K with normalizing coefficients D₁...DK. Compensation of the errors of the output signal and unaccounted disturbances implemented in IB 5₁...5K with normalizing coefficients And₁...IR.

Next output PY 3, the signaling information is supplied for informational inputs CFLs 9.

According to the algorithm presented in figure 2, in CFLs 9 calculated values of control signals for all loops and defined values of the resulting coefficients. While CFLs 9, which implements the control logic uses conditional rules laid down in the software, to convert an array of input signals in control.

Control signals from outputs of CFLs 9 served on Cmvh 1, Cmwyj 2, PN 3, PB 4 and IDB 10, and can preconfigurated coefficients PB 4, And₁...IR and D₁...DC block 10.

Receiving the control signals, Cmvh 1, Cmwyj 2 can reconfigure the structure of the compounds.

Receiving the control signals, RU 3 generates an output signal u(kt). Output PY 3 signal u(kt) is fed to the input of the HMB 11, where the signal is converted into a form understood PP 12.

The mechanism back the connection is implemented by the signal y(kt) output UO 12 CMS 1.

Since the system consists of many interconnected with Cmvh 1, Cmwyj 2 blocks IDK 7₁...7K, the number and structure of relationships can vary within the existing tasks. Due to the nonlinearity of the function (8) and a large number of adjustable coefficients, it is possible to obtain a flexible and configurable system structure.

It is obvious that for the degenerate case of absence of the second term in (8) we can obtain the expression:

$$u_{\sum }={\displaystyle \underset{i}{\sum }{\lambda }_i{u}_i.\left(9\right)}$$

For the degenerate case of absence of the first term in (8) we can obtain the expression:

$$u_{\sum }={\displaystyle \underset{i,j}{\sum }{\lambda }_{ij}{u}_i\left(u_j\right)}.\text{what is}\left(10\right)$$

The structure of the interconnections corresponding to the formula (9)shown in figa - parallel structure.

The structure of the interconnections corresponding to the formula (10)shown in figb sequential structure.

Some of the possible implementation of the combined structures of the proposed system is shown in figv, ,

The proposed system with multiple control loops can be implemented on the following elements:

switching matrix can be implemented on the FPGA elements [9] (programmable logic integrated circuits) or digital logic[7, 9, 10];

- solver, proportional, integrating, differentiating and execution units can be implemented on specialized analog and/or digital devices designed to control the desired physical parameters of the system (for example, to control the frequency of multifunctional sensors, including the ADC and/or DAC) [7, 8, 10, 11];

controller functional logic can be configured FPGA-based firm ALTERA [9], which contains an embedded 32-bit processor on a chip NIOS, or on the basis of a multi function you who echointegerarray data processing system, including high performance h-compatible kernel and Flash memory [12].

The proposed method is, in essence, can be called fuzzy PID controller parametric type. The control algorithm consists of several hierarchical levels, including direct control of the Executive block (ISB) through solver (RU), accumulation of samples of the previous values of the system parameters and calculating the current settings integramouse-differentiating clusters (IDK), and development of predictive behavior of the system based on the long history and calculation of control parameters PN and IDK for this case.

Thus, it is possible to improve the functionality of the system using the methods of multi-level logic and adaptive algorithms, operating with incomplete and inaccurate data. One system used a combination of several methods on the basis of a base of fuzzy rules and the use of the principles of training the neural network.

The improved functionality is achieved due to the fact that for the implementation of adaptive control by the PID act added functional logic controller; integrating and differentiating blocks, each of which has its own normalizing factor, appointed by the controller of the functional logic, in the United integrirovannymii clusters, included in the ID-block; added solver carrying values of the levels of the regulated signals received from ID-block, the controller functional logic for making decisions about the need for correction of the parameters of the managed system. Cascade connection between ID-blocks, which are implemented through the switching matrix, under the control of the controller functional logic, give the opportunity to get a large number of options interface network interconnections to implement the adaptive structure of the managed system.

The proposed method and system are particularly effective for systems management with a significant level of a priori uncertainty of the parameters with the purpose of providing additional protection in terms of special effects and intentional interference.

Sources of information

1. Ball, Stuart R. the Analog interfaces of the microcontroller. / M: Publishing house "Dodeca-XX1", 2007. - 360 C.

2. Olsson, Digital automation systems and control / Olsson,, Piani j. // SPb.: Nevsky dialect, 2001. - S. 557

3. RF patent for the invention №2157558 "Supervisory proportional-integral-differential controller". / Dialects A.A., Kuzmichev E.V., Speak, S.A., 2000.

4. RF patent for the invention №2419122 "Self-tuning PID controller". / Alex the wood A.G., Palenov M.V., 2011.

5. Mandel, I.D. Cluster analysis. / M.: Finance and statistics, 1988. - 176 S.

6. Fowler, M. Refactoring: improving the existing code. / Lane. from English. - SPb: Simbolos, 2003. - 432 S.

7. http://www.analog.com/ru/allProducts.html - electronic Components company "Analog Devices".

8. http://russia.maxim-ic.com/products/ - electronic Components company "MAXIM".

9. http://www.altera.ru/cgi-bin/go?38 - electronic Components company "ALTERA".

10. http://www.ti.com/ww/ru/ - electronic Components company "Texas Instruments".

11. http://www.murata.com/ - electronic Components company "Murata Manufacturing Company, Ltd.

12. http://www.silabs.com/products/pages/default.aspx - electronic Components company "Silicon Labs".

13. Buscarse V.A. Theory of automatic control systems / Vasetsky, Appatow. - SPb.: Profession, 2004. - 752 S.

14. Gostev VI Fuzzy controllers in automatic control systems. / K.: Radiometer, 2008. - 972 S.

15. Zhdanov A.A. Possibilities of using technology deterministic chaos in systems of Autonomous adaptive control / ADV, Aestuans // M: ISP RAS, 2001 - S-179. (Proceedings of the Institute for system programming).

16. Kim D.P. Theory of automatic control. Vol.2. Multidimensional, nonlinear, optimal and adaptive systems: Textbook. the allowance. / M.: FIZMATLIT, 2004. - 464 S.

17. Terekhov V.A. Neural network control system. / Terekhov V.A., Efimov D.V. Tokin IU, Antonov V.N. // St. Petersburg: St. Petersburg state University, 1999.

18. Chkhartishvili A.G. game-Theoretic model of information management. / M: CJSC "PMSOFT", 2004. - 227 C.

1. Method of adaptive control on the basis of the law of the PID, which consists in the following steps:
using a priori parameter values configuration of a managed system (MOUSTACHE) for the specified operating conditions CONDITION, determine acceptable levels of criticality deviations of the relevant configuration settings for the CA in the form of logical functions defined on respective sets of deviations of the values of the configuration parameters of the CA, according to the possible levels of the deviations of the input signal from the corresponding values of the matrix of decision making;
the obtained values of acceptable levels of criticality deviations configuration CONDITION is used to install one correspondence between the parameter configuration CONDITION and level of criticality reject;
the values of the matrix of decision making compute the error variance of the values of the configuration parameters MUSTACHE;
if the calculated error variance values of the configuration parameters CONDITION correspond to a valid severity of deviations configuration settings MUSTACHE, and retain the previous settings configuration MUSTACHE and conduct initial refactoring is receiving a set of values of a matrix of decision-making;
if the calculated error variance values of the configuration parameters CONDITION does not correspond to a valid severity of deviations configuration settings MUSTACHE, remember this event and the calculated error variance values of the configuration parameters MUSTACHE;
next, perform the adaptation values of the configuration parameters of the CA to the identified deviations functioning CONDITION by assigning a controlling influence on US on the basis of selecting values from a matrix of decision making, when it set up a correspondence between current and valid configuration values MUSTACHE;
if the identified missing and/or incorrect settings that you have defined the configuration of the CA, perform the refactoring of the content of the set of values of a matrix of decision making for each of the configuration parameters CONDITION for the specified operating conditions CONDITION by adding the identified missing values and/or changes to incorrect values of the matrix of decision making;
then retain the previous settings configuration MUSTACHE, appointed to manage the impact on US and spend refactoring content of the set of values of a matrix of decision making by replacing the previous values of the configuration parameters of the CA current values of configuration parameters MUSTACHE;
repeat the above steps, beginning the phase determination of the severity of the deviations of the configuration parameters CONDITION for the specified operating conditions MUSTACHE.

2. System for adaptive control of the PID act, contains a switching matrix inputs (Kmwh) and outputs (Cmwyj), connected in series solver, the Executive unit and the managed object whose output, which is the output device, connected to the second signal input Kmwh, the first signal input which is the input of the controlled signal; a controller functional logic (CFL), is proportional to the unit and integramouse-differentiating unit (IDB), including To integramouse-differentiating clusters (IDK), each of which contains at integrating and differentiating unit with its own normalizing coefficients, and To pairs of signal inputs IDB are pairs of signal inputs corresponding IDK where the first signal input IDK signal is input corresponding to the integrating unit and the second signal input IDK is the signal input of the corresponding differentiating unit; output pairs IDB are pairs of outputs corresponding IDK, from which the first output is the corresponding output of the integrating unit, and the second output is the corresponding output of the differentiating unit; a first output Cmvh connected with the signal input is proportional to the unit, the output of which is connected to the first signal input is Psych, the first output of which is connected with the first signal input of a casting device, the other pairs To the signal inputs of which are connected with corresponding To pairs of outputs Cmwyj; the remaining pairs of outputs Cmvh connected to respective pairs To the signal inputs IDB TO pairs of outputs which are connected with the respective pairs To the signal inputs of Cmwyj, group outputs, which by means of the signal bus is connected with a group of signal inputs Cmvh; the first output of CFLs is connected with the control input Kmwh, the second output with the control input is proportional to the block, the third output - control inputs To each of the integrating and differentiating blocks; the fourth output - managing input Cmwyj, and the remaining outputs CFLs connected to respective control inputs of a casting device, the information outputs of which are connected to the appropriate information inputs CFLs.