Administration of complex technological processes and productions

Mathematical modeling and process control in industrial automation systems using complex-valued neural networks

V. N. Meshcheryakov, O. V. Meshcheryakova, P. V. Saraev

Lipetsk State Technical University

Abstract: The paper considers a multi-valued threshold logic, which is the base of multi-valued neurons, discusses the advantages of complex-valued neural networks. The complex-valued activation function is defined for the multi-valued neuron and the backpropagation learning algorithms for the multi-valued neuron and complex-valued neural networks are considered. Further, a method of applying complex-valued neural networks for modeling processes is proposed.

Keywords: mathematical modeling, complex-valued neural networks, neurocontrol.

UDC: 004.89