Comparative analysis of fuzzy neural network structures for formation model of electricity consumption in the traction power supply system

Nowadays fuzzy neural networks are widely used for modeling of complex industrial processes. The paper considers the application of fuzzy logic to generate a mathematical model of electricity consumption in rail transport for example, traction substation Dorogino. Algorithm for choice of the structure of fuzzy neural network including the species and number of membership functions for input and the number of training cycles is presented. Comparative analysis of the structures by evaluating the mean square error is made.