Diveev

The paper compares two methods of symbolic regression, namely a method of genetic programming and a variational method of genetic programming. The comparison is based on the computing experiment, which solves a problem of control system synthesis for a model of nonlinear control object, describing the interaction of the two systems i.e. "predator" and "victim". For the purity of the experiment the genetic algorithms parameters in both methods were identical. For variational genetic programming there was selected a trivial basic solution in the form of the sum of input variables products for custom settings. This basic solution is always chosen with a lack of meaningful task analysis. The methods are compared both in terms of the speed for problem solving and in terms of the quality of the achieved control.

The authors consider the identification problem for a chemical reaction model. The mathematical model in the form of a system of ordinary differential equations is developed on the basis of the given experimental data on the variation in time index of the concentration of substances that interact in the chemical reaction. To construct this model which clarifies the law of mass action a network operator is used. As a result of application of the network operator method, clarification of the law as a product of polynomials of the concentration of substances that interact in the chemical singlet oxygen reaction is obtained.

 The authors formulate a problem of control systems synthesis for dynamic objects with random initial values. It was proposed for the synthesis to modify functionals of quality to form sum of values with weights for each initial value. Weights determine probabilities of initial values.  It was proposed to use the numerical network operator technique for solving the synthesis problem. An example of control system synthesis for an aircraft at the descend stage is presented in the article.