The Research About Gear Box Failure Diagnosis Based On The LVQ Neural Network

In recent years, with the development of modern production, the mechanical equipment's security and reliability problems have become increasingly prominent. The equipment glitches appear may result in paralysis of the whole system, in order to improve equipment reliability, reduce equipment forced outages, reduce maintenance costs, and extend the service life of equipment, the mechanical equipment's fault diagnosis technology have been put more and more attention. Fault diagnosis techniques is a diagnostic technique that developed with the development of modern industrial mass production equipment, and one of the key technologies that is the safe and reliable operation of large machinery and equipment, but also is the basis for which a variety of automated systems and general mechanical system efficiency and reliability predictive management and predictive maintenance. Therefore, the research of fault diagnosis on mechanical equipment is of great significance.LVQ neural network transfer the function through the competitive hidden layer, and depend on the target's category classification of the input vector combinations. The competitive layer can learn the classification of the input vector. The linear layer changes the classification of information from the competitive layer into the categories defined by the user. The advantage of LVQ neural network: it not only to classify the linear input data, but also deal with multi-dimensional, even with interference of noisy data. Of course, LVQ neural networks has some inadequacies: The most serious problem is no ability to explain their reasoning process and reasoning basis; Can not make the necessary inquiry to the user, and the time when the data is not sufficient, neural networks can not work, turn all the issues features into digital and turn all the reasoning into numerical computing, the result is bound to lose information; theory and learning algorithms should be further improved and enhanced.This paper first introduces the theory of artificial neural network, and structure the BP network model based on the theory of BP network and the LVQ network model based on the theory of LVQ network. In the instance of Gearbox fault diagnosis, respectively using BP neural network and LVQ neural network to diagnose the gear case's fault, then compare the results of the Gearbox fault diagnosis and obtain that the LVQ neural network have more advantages in fault diagnosis. Finally, sum up the work, and propose future research directions.