Control law and pseudo neural networks synthesized by evolutionary symbolic regression technique

Abstract

This research deals with synthesis of final complex expressions by means of an evolutionary symbolic regression technique-analytic programming (AP)for novel approach to classification and system control. In the first case, classification technique-pseudo neural network is synthesized, i. e. relation between inputs and outputs created. The inspiration came from classical artificial neural networks where such a relation between inputs and outputs is based on the mathematical transfer functions and optimized numerical weights. AP will synthesize a whole expression at once. The latter case, the AP will create chaotic controller that secures the stabilization of stable state and high periodic orbit-oscillations between several values of discrete chaotic system. Both cases will produce a mathematical relation with several inputs, the latter case uses several historical values from the time series. For experimentation, Differential Evolution (DE) for the main procedure and also for meta-evolution version of analytic programming (AP) was used.