Application Research on Fault Diagnosis Based on Improved Wavelet Neural Network

摘要

Aluminum electrolysis is a nonlinear, multi-couplings, time-variable and large time-delay industrial process system. The paper puts forward the fault diagnoisis method of improved wavelet Elman neural network, which firstly simplifies the input of network with the method of principal component analysis, secondly, the weights, as well as scale factor and shift factor of the wavelet function are optimized by use of the wavelet Elman network which is optimized by improved particle swarm algorithm. Then it is verified by the simulation. The simulation results show that the method can precisely forecast the aluminium electrolysis equipment faults and improve the production and quality of aluminum.