The Application of BP Neural Network on Mechanical Failure Classification

Abstract

Based on the fuzzy classifying approach, the paper puts forwards a diagnosis algorithm of Back-propagation Neural Network. For some complexity environments, the traditional Backpropagation Neural Network has some limitations on classification. The paper applies fuzzy model on Neural Network structure, by using classifying variance and energy function to adjust the convergence of the Neural Network. With the improved nonlinear mapping property, the diagnostic processing shows perfect results with identifying ratio of 100 percent, while the traditional method is 65 percent only. Keywords—Classification, Neural network, Diagnosis, Back propagation