Application of modified genetic programming algorithm for identification of mathematical models through the expansion of the training set by neural network

The concept of mathematical identification, its scope and stages of implementation are considered. The methods of identification of mathematical models: regression analysis, harmonic analysis, group method of data handling, genetic programming are analyzed. The restriction of the use of genetic programming method for the identification of the mathematical model of unexplored process in the presence of the noise component in the experimental data is studied. Proposes a modification of the method of genetic programming using the method of pre-approximation and expanding the training set by artificial neural network. The interfaces of the developed software product and the test results of the proposed method are presented.