Genetic algorithm based variable-structure neural network and its industrial application

Abstract

This paper presents a neural network model with a variable structure, which is trained by an improved genetic algorithm (GA). The proposed variable-structure neural network (VSNN) consists of a neural network with link switches (NNLS) and a network switch controller (NSC). In the NNLS, switches in its links between the hidden and output layers are introduced. By introducing the NSC to control the switches in the NNLS, the proposed neural network can model different input patterns with variable network structures. The proposed network gives better results and increased learning ability than conventional feed-forward neural networks. An industrial application on short-term load forecasting in Hong Kong is given to illustrate the merits of the proposed network.