Fuzzy Neural Network Expert Fault Diagnosis System Based On LM-BP Optimization Algorithm

In recent years,with the development of the power electronics technology and the theory of numerical control system,digital servo system with semiconductor converter technology is widely used in various types of military and civilian equipment.Modern servo drive system is composed of complex structure,so its fault is unavoidable when it is running.But the end users can't correctly diagnose servo system timely because they have not enough relating knowledge.Traditional fault diagnosis methods,such as fault tree diagnosis method,if the diagnosis of objects is more complex,the establishment of fault tree will be difficult;mathematical modeling analysis fault diagnosis method for non-linear complex system is helpless;pattern recognition fault diagnosis must be pre-accumulated a large number of fault diagnostic knowledge.In view of these traditional methods,it is difficult to carry out comprehensive fault diagnosis of a nonlinear complex system such as AC servo drive system.In this paper,expert system,fuzzy logic and artificial neural network method based on Levenberg-Marquardt BP(LM-BP)optimized algorithm,three kinds of artificial intelligence technology are combined with B&R AC servo drive system fault diagnosis of the actual situation.This paper is to study fault diagnosis system's construction and implementation technology of B&R AC servo drive system.Three methods of mutual integration,learn from each other,and play their own characteristics.Fuzzy reasoning overcomes the interference caused by the uncertainty,inaccuracy,and noise of the process itself.The membership degree is used to describe the uncertain relationship between fault and sign due to ambiguity.The expert fault diagnosis system is like the left brain of human beings,using the rules of clear expression to arrive at the correct conclusion of the logic,and the reasonable explanation of the reasoning process.The neural network is similar to the human's right brain.In the decision of the problem,the neural network is better than expert fault diagnosis system,and they are two complementary methods.