Forecasting System Based On Engine Failure Of The Neural Network And Sound Intensity

After compared the selection of engine characteristic parameters and main forecast modules, detailed researched the way of constructing engine fault forecast system based on sound intensity analysis and neural network (NN), designed its software system using VB and MATLAB computer languages.The thesis firstly analyzed the occurring mechanism of normal engine faults which represent as abnormal noise, investigated practical use of sound intensity technology in engineering, discussed the feasibility which use sound density frequency analysis to forecast engine fault. Then compared the main forecast modules, according to non-stability of engine fault appearance, chose BP NN module to forecast the normal fault of engine. Forecast steps firstly begin at recognizing the main noise of engine, then analyzing the sound intensity frequency and chart characters selectively. The thesis discussed the BP network structure, characteristic parameters and learn methods using practical examples in the phrase of building the NN module, then forecasted the developing trend of engine's fault using the NN module which was learned well, proposed the maintain measures accordingly. Completely wrote and compiled the software of fault forecast system using high lever computer languages.In the end, through running the software and instances, validated the feasibility and correctness which using sound intensity analysis and neural network to forecast engine's main fault.