Method for multialternative optimisation of automation modules of structural synthesis of mechatronic modular robots

FIELD: machine building.

SUBSTANCE: method for multialternative optimisation of automation modules of structural synthesis of mechatronic modular robots is proposed, in which at performance of synthesis of the multiinvariant model structure of mechatronic modular robots, and further fixation of obtained optimum solutions, a variety of design elements is considered and corresponding alternative variables are entered by presenting discrete numbers corresponding to those elements in binary notation; after that, the number of modules combined in one robot, mainly without distinct structure are marked, and connection of every new module is provided to earlier assembled ones along the chosen direction and coupling of its first interface platform is performed to one of the free ones on any other structural members occupying the closest extreme position in this or that row; after that, alternative variables are entered; at that, for optimisation structural synthesis there chosen are values of alternative variables $$\overline{x_1^{*},x_{41n}^{*}}$$ providing maximum value of function f.

EFFECT: enhancement of a synthesis process; improvement of operating reliability of mechatronic modular robots.

2 cl, 4 dwg

The invention relates to the field of engineering, namely robotics, and can be used to create Megatron-modular robots.

One of the most important and promising directions of development of modern robotics is associated with the development of a new class of devices - multilink Megatron-modular robots with adaptive structure. Structural synthesis when designing reconfigurable Megatron - modular robots is considered simultaneous, automated solution two choice tasks: order modular Assembly and configuration options priori periodic law of change of generalized coordinates (y, z)that defines the algorithm for motion control.

There is a method of multi-optimization models for automation of structural synthesis Megatron-modular robots, which consists in conducting the synthesis, structure mnogovershinnoe model Megatron-modular robots, and the subsequent fixation of the obtained optimal solutions (IGOR Makarov, V.M. Lokhin, S. Manko, BTW, Novels, M.V. Kadochnikov. It, "processing Technology of knowledge management tasks offline Megatron-modular reconfigurable robots", Annex "Information technology" No. 8, M., "New technologies", 2010, p.3-7, 14-prototype).

This method multiple-choice optimization mod is lei automation of structural synthesis Megatron-modular robots is memorizing specific provisions of individual modules to resolve targets.

The disadvantages of this method is its considerable complexity, low efficiency orientation in the environment reconfigurable mechatronic devices, mainly Megatron-modular robots.

The objective of the proposed technical solution is to eliminate these disadvantages and to create multi-way optimization models automation of structural synthesis Megatron-modular robots, which will speed up the process of synthesis, as well as increase the efficiency of orientation in the environment and reliability created mechatronic devices, mainly Megatron-modular robots.

The solution of this problem is achieved by the fact that in the proposed method multi-optimization models for automation of structural synthesis Megatron-modular robots, according to the invention, when carrying out synthesis patterns mnogovershinnoe model Megatron-modular robots, consisting of at least two connected between an identical modules, preferably, two or more primary and again with him mating/s, has a front-end platform for docking, and the number of modules to be merged in the above-mentioned robot, op is Adelino from the relation: n=1,N, where: n is the number of modules to be merged into one robot, determined from the relation n=1+x1+2x2+4x3+8x4, where: x1, x4=1.0 is the number of interface pads on the module, N≤16 - maximum number of modules that can be combined into one robot, while pairing each new module with the previously collected/and implemented along the selected direction and provided with a coupling its first front-end sites with one free on any other structural elements occupying the nearest extreme position in a particular row, and the interface pad of each module is configured to dock with the same sites, at least four diametrically opposite directions, and the subsequent fixing of the obtained optimal solutions consider a number of design elements and enter the appropriate alternative variables by representing digital numbers corresponding to these elements, in binary terms, then indicate the number of modules to be merged into one robot, mostly, without a distinct structure, and provide a pair of each new module with the previously collected along the selected direction and joining his first front-end sites with one free on any other structural elements occupying the nearest extreme the position at which one or another row, then enter the alternative variables to describe the parameters of the periodic law as follows:

Angle=A+Bsin(ωt+φ),

where: A - the value of the generalized coordinates, relative to which there is a periodic motion;

In - amplitude periodic oscillations of the generalized coordinates; total value of a+b must not exceed the maximum permissible variation of the generalized coordinates of the module;

φ is the phase shift of the periodic motion.

thus, for optimization of structural synthesis of choosing alternative values of the variables$$\stackrel{}{x_1^{*},x_{41n}^{*}}$$providing the maximum value of the function f:

$$f=\frac{{\left[y\left(\stackrel{}{x_1,x_{41n}}\right)\right]}^2+{\left[z\left(\stackrel{}{x_1,x_{41n}}\right)\right]}^2}{N\left(\stackrel{}{x_1,x_{4n}}\right)N_c\left(\stackrel{}{x_{10},x_{41n}}\right)}\to \max$$

when the constraint n=1, N

$$\left|A_1\left(\stackrel{}{x_{10},x_{12n}}\right)+B_1\left(\stackrel{}{x_{14n},z_{17n}}\right)\right|\le y^{\max },$$

$$|A_2\left(\stackrel{}{x_{26},x_{29n}}\right)+B_2\left(\stackrel{}{x_{30n},z_{33n}}\right)|\le z^{\max }$$

$$\stackrel{}{x_1,x_{41n}}=\{\begin{array}{c}\mathrm{1,}\\ 0.\end{array}$$

where: y^max, z^max- maximum allowable deviation of the generalized coordinates of the module relative to its zero value, for finding the maximum value of the function f, using a randomized algorithm multi-optimization.

In the use case method, for finding the maximum value of the function f, a randomized algorithm for multi-optimization complement another level within the controlled particle swarm optimization.

The invention is illustrated by drawings, where figure 1 shows the individual Megatron-modular robots with free front-end platforms, figure 2 - Megatron-modular robot, consisting of several modules, interconnected by a free interface pads and forming the shape of the polygon in figure 3 - Megatron-modular robot, consisting of several modules, interconnected by a free interface Playground and forming a shape in the form of a square, figure 4 - Megatron-modular robot, consisting of several modules, interconnected by a free interface pads and forming the shape of a rectangle.

In the drawings under item 1 included as a separate Megatron-modular robot that consists of a single module, pos.2 - free interface pad, 3 - front Playground, used for mating with another individual Megatron-modular robot that consists of a single module, pos.4 - Megatron-modular robot, consisting of several modules 1, interconnected by a free front-end sites 2.

The proposed method can be implemented as follows.

Consider a number of design elements and enter the appropriate alternative variables by representing digital numbers corresponding to these elements, in binary terms.

Denote the number of modules 1 may be merged into one Megatron-modular robot 4, without a distinct structure,$$n=\stackrel{}{\mathrm{1,}N}$$Then in binary terms obtained when N≤16, where: N is the number of sides, n is the number of possible iterations.

n=1+x₁+2x₂+4x₃+8x₄,

where$$\stackrel{}{msub>x1,x_4}=\{\begin{array}{c}\mathrm{1,}\\ 0.\end{array}$$

When modular Assembly robot 4 believe that the pairing of each new module 1 with previously collected is performed along the selected direction and is provided by coupling its first interface pad 2 with one of the available similar interface pads 2 on any other modules 1, as structural elements, occupying the nearest extreme position in a particular row.

The highlights of this algorithm mainly as ASB. Description of the Assembly procedure lead to the indication of direction and the attachment of another element using the SSA algorithm.

In the direction of the docking of the n-th module of pet take four values of n_cm=1 - North, n_cm=2 - East, n_cm=3 - South, n_cm=4 - West and represent through alternative variables:

n_cm.n=1+x_5n+2x_6n,

where$$n=\stackrel{}{\mathrm{1,}N}$$,$$x_{5n},x_{6n}=\{\begin{array}{c}\mathrm{1,}\\ 0.\end{array}$$

The number of sites selected for docking of the n - th module in binary, write in the following form:

n_cm.n=1+x_7n+2x_8n+4x_9n,

where$$n=\stackrel{}{\mathrm{2,}N}$$,$$\stackrel{}{x_{7n},x_{9n}}=\{\begin{array}{c}\mathrm{1,}\\ 0.\end{array}$$

Alternative variables to describe the parameters of the periodic law is administered as follows:

Angle=A+Bsin(ωt+φ),

where: A - the value of the generalized coordinates, relative to which there is a periodic motion;

In - amplitude periodic oscillations of the generalized coordinates; total value of |A|+|B| must not exceed the maximum permissible variation of the generalized coordinates of the module;

φ is the phase shift of the periodic motion.

The settings of this law define the control algorithms synthesized Megatron-modular design. The specified parameters x is characterized by discrete values, with appropriate numerical numbers in the range N≤16.

Then for optimization of structural synthesis of choosing alternative values of the variables$$\stackrel{}{x_1^{*},x_{41n}^{*}}$$providing the maximum value of the function.

$$f=\frac{{\left[y\left(\stackrel{}{x_1,x_{41n}}\right)\right]}^2+{\left[z\left(\stackrel{}{x_1,x_{41n}}\right)\right]}^2}{N\left(\stackrel{}{x_1,x_{4n}}\right)N_c\left(\stackrel{}{x_{10},x_{41n}}\right)}\to \max$$

when the constraint n=1, N

$$\left|A_1\left(\stackrel{}{x_{10},x_{12n}}\right)+B_1\left(\stackrel{}{x_{14n},z_{17n}}\right)\right|\le y^{\max },$$

$$\left|A_2\left(\stackrel{}{x_{26},x_{29n}}\right)+B_2\left(\stackrel{}{x_{30n},z_{33n}}\right)\right|\le z^{\max }$$

$$\stackrel{}{x_1,x_{41n}}=\{\begin{array}{c}\mathrm{1,}\\ 0.\end{array}$$

where y^max, z^mx - maximum allowable deviation of the generalized coordinates of the module relative to its zero value.

To find the maximum value of the function, hdaci use a randomized algorithm multi-optimization, which is complementary to another level within the controlled particle swarm optimization.

For synchronization procedures of the method of particle swarm optimization and variational procedures multi-optimization at each step is controlled by selection of the particle to update the rate of change of coordinates, which is carried out using a randomized scheme. To this end introduce random discrete value m, which takes the value m=1,M with probability pn. In the first step, get:

$$p_n^1=\frac{1}{N}{\forall }_n=\stackrel{}{\mathrm{1,}N}$$.

Next, change the values of$$p_n^k$$provided$${\displaystyle {\sum }_{n=1}^M{p}_n^{{\nu }_k}=1}$$as follows. About Radelet value of the random variable $$\stackrel{}{n}$$. Let$$\stackrel{}{n}=\nu$$. Then the rate of change of coordinates (k+1)-th step are calculated:

$${\nu }_{mn}^{r+1}=\{\begin{array}{c}{\nu }_{mn}^r,\forall n=\stackrel{}{\mathrm{1,}N},n\ne \nu ,\\ p_{B_{mn}}^{r+1}[q_{z_{mn}}^{r}x\left({\nabla }_{1m}F\right)-p_{z_{mn}}^{r}x\left(-{\Delta }_{1mn}F\right),\\ n=\nu \end{array}$$

and the value of probability p_n:

$$p_n^{k+1}=\{\begin{array}{}\frac{p_n^k}{1+{\epsilon }^{k+1}}\forall n=\stackrel{}{\mathrm{1,}N},n\ne \nu ,\\ \frac{p_n^k+{\epsilon }^{k+1}}{1+{\epsilon }^{k+1}},n=\nu .\end{array}$$

The value of ε>0 determines the degree of recordnet movement of the ν-th particle in the direction of the extremum of the function being optimized.

The use of the proposed technical solution will allow the synthesis of structure mnogovershinnoe model Megatron-modular robots, and subsequent recording of the obtained optimal solutions, with the consequent increase of the number of possible iterations Megatron-modular robot with a significant reduction in synthesis time.

1. The multi-way optimization models automation of structural synthesis Megatron-modular robots, characterized by the fact that when conducting structure synthesis mnogovershinnoe model Megatron-modular robots, consisting of at least two connected between the Wallpaper identical modules, preferably, two or more primary and again with him mating/s, has a front-end platform for docking, and the number of modules to be merged in the above-mentioned robot, determined from the relation: n=1,N, where n is the number of modules to be merged into one robot, determined from the relation n=1+x1+2x2+4x3+8x4, where:x1, x4=1.0 is the number of interface pads on the module, N≤16 - maximum number of modules that can be combined into one robot, with each pair the new module with the previously collected/and implemented along the selected direction and provided with a coupling its first front-end sites with one free on any other structural elements occupying the nearest extreme position in a particular row, and front-end areas of each module is configured to dock with the same sites, at least four diametrically opposite directions, and the subsequent fixing of the obtained optimal solutions consider a number of design elements and enter the appropriate alternative variables by representing digital numbers corresponding to these elements, in binary terms, then indicate the number of modules to be merged into one robot mainly, without a distinct structure, and provide each pair but the second module with the previously collected along the selected direction and joining his first front-end sites with one free on any other structural elements, occupying the nearest extreme position in a particular row, then enter the alternative variables to describe the parameters of the periodic law as follows:
Angle=A+Bsin(ωt+φ),
where a is the value of the generalized coordinates, relative to which there is a periodic motion;
In - amplitude periodic oscillations of the generalized coordinates;
total value of a+b must not exceed the maximum permissible variation of the generalized coordinates of the module;
φ is the phase offset of a periodic motion,
for optimization of structural synthesis of choosing alternative values of the variables$$\stackrel{}{x_1^{*},x_{41n}^{*}}$$providing the maximum value of the function f:

when the constraint n=1, N
$$\left|A_1\left(\stackrel{}{x_{10},x_{12n}}\right)+B_1\left(\stackrel{}{x_{14n},z_{17n}}\right)\right|\le y^{\max }$$,
$$\left|A_2\left(\stackrel{}{x_{26},x_{29n}}\right)+B_2\left(\stackrel{}{x_{30n},z_{33n}}\right)\right|\le z^{\max }$$
$$\stackrel{}{x_1,x_{41n}}=\{\begin{array}{c}\mathrm{1,}\\ 0.\end{array}$$
where y^max, z^max- maximum allowable deviation of the generalized coordinates of the module relative to its zero value, for finding the maximum value of the function f using a randomized algorithm multi-optimization.

2. The method according to claim 1, characterized in that to find the maximum value of the function f is a randomized algorithm multi-optimization complement another level within the controlled particle swarm optimization.