| Firstly the thesis has studied some basic methods of mechanical trouble diagnosed,and especially made a deep research on the methods of back and forth machine's trouble diagnoses.Against the important mechanic in project,the thesis discussed engine classification characteristic and methods of trouble in detail.Comparing the advantages and disadvantages between traditional diagnoses and intelligence diagnoses,the thesis studied and summarized emphatically the method of artificial neural network trouble diagnosing which is being applied at the field of trouble diagnosing widely in recent year.The engine as the representative of back and forth machine, the trouble demonstrates characteristics of complexity and multiplicity.It's difficult for us to discern the trouble by using traditional diagnosis method, and the diagnose accuracy is very low. With development and application on of artificial intelligence technology,more and more researches applied it at trouble diagnosis field and have acquired the result better than traditional methods.The thesis would discuss the application of artificial neural network at the field of trouble diagnosis in detail, especially regarding engines as research object researched the method of neural network on trouble diagnose of engine deeply.And emphatically introduced mathematic mathematics principle of BP network, explored the setting-up of network model structure, the parameter choose of network, network training way, network treatment methods etc,at last,summarized several kinds improved method of BP.By applying VC++ and MATLAB,the author sets up software of engine trouble diagnosis. The database's operation is based on SQL Server 2000,achieving preliminary intelligence diagnosis. Finally, the thesis took lubrication system as an example,made the trouble diagnosis in order to check the system, and the result showed that the engine diagnosis based on neural network was executable and correctly.
|
PDF Full Text Request