I.C. Engine Misfire Fault Diagnosis Based on Singular Value Theory and Neural Networks
Hua Wang · Transactions of Csice · 2003
The explosive noise signal of multicylinder of I.C. Engine is decomposed with Station Wavelet Transform. Parts of decomposition are used to compose the Cover Matrix. The Cover Matrix enhanced information of cylinder explosive noise signal at frequency band of cylinder explosive. Singular value is extracted as to feature of cylinder explosive noise signal from Cover Matrix with singular value theory. The improved BP neural network was trained with singular value as the ideal input and codes of fault states as the ideal output. Research indicates that this method can be used to diagnosis misfire fault of multicylinder of I.C. Engine effectively.