基于神经网络的汽车变速器可靠性研究-机械工程专业论文.docxVIP

ABSTRACTBeing ABSTRACT Being one of the important pans of automobile assemblies，the reliabilit、，of transmission is the main factor directly affecting automobile security．As the basis of reliability analysis is to gather and analyze large mount of failures data。the key is how to built the models of reliability and get estimated parameter．Due to its unique structure and information processing，artificial neural network has been successfully applied in many practical fields and solved many problems that can’t be solved with the traditional computer easily． Reliability analysis based on artificial neural network will introduce the computer learning to distributing processing．so that the level of the automation and intelligence of distributing recognizing will be enhance．This application will also create a new way for reliability theory to solve engineering and technology problems，and will develop and complete reliability engineering regarding of reparability, which will provide great guidance to theory perfection and science innovation． IIl tIlis dissertation．Foton BJl 028 mini．truck is used as a model to analyze the reliability of transmission in terms of the method of recognizing reliable model and estimation of parameter．As all maintainable system，the failures and mean life oftransmission accord with Weibull distribution．1 will use Back-Propagation Network to recognize reliable model and use Madaline to estimate the parameter．Improved BP arithmetic Can be used because it Can self-adjust the learning rate on the basis of the structure of BP network． In the essay,1 will compare and analyze the characteristics and applications of several reliable models．which deal with Wbibull distribution，which deals with the complicated equipments．Due to the advantage of ANN on distributing recognizing，I designed the model-recognizing system and show how it works by the parameter estimate of index distributing and Weibull distribution．Reliable model m recognizing recognizing system ca