Application Analysis of Structural Equation Model Based on BP Neural Network Algorithm in Fault Diagnosis of Power Plant Boilers

Abstract

As the core equipment of combustion, the safe operation of the boiler is of vital importance. Due to the complex structure of the boiler, damage, abrasion, acid gas corrosion and improper operation will all cause faults. In order to effectively avoid faults, a multi-dimensional BP neural network method is used for boiler fault diagnosis modeling, in which the BP neural network adopts multi-dimensional structure, the input layer adopts fuzzy mathematics method to quantify the operation parameters, and a multi-dimensional BP neural network model is established through the correlation between parameters and between parameters and fault types. The experimental results show that the BP neural network fully inherits the advantages of wavelet transform and neural network. The method has good fault diagnosis ability and is obviously superior to wavelet neural network in the accuracy of fault diagnosis.