Diesel Engine Fault Diagnosis Based On KPCA And Information Entropy

In science and technology today, people have become increasingly demanding mechanicalequipment. Reciprocating diesel engine as the most common mechanical equipment, is widelyused in various fields of automobile, aircraft, ships and industrial and agricultural. However,due to the complexity of its structure and Diesel poor working conditions and other factors,led to the failure of multiple diesel and diversity. How to detect and effective and rapiddiagnosis and treatment failure has become a tireless force scholars and research..In this paper, R6105AZLD diesel for the study, to obtain the original signal and the signalacquisition system failure settings. Based on this proposed diesel engine fault diagnosismethod based on kernel principal component analysis and information entropy. Waveletpacket noise inherited wavelet analysis of more debate and more features, but also overcomethe wavelet noise reduction processing for high-frequency signal is weak shortcomings, tobetter improve the time-frequency resolution. Through improved and redesigned kernel usingkernel principal component analysis in terms of the unique advantages of nonlinear signalprocessing, to map the nonlinear coupling between noise wavelet packet data into a highdimensional feature space, linear high-dimensional feature space PCA; short for dieselvibration signal, a signal from a quantitative description of the method of starting the use ofdelay embedded depression, baseline drift signal restoration techniques, respectively, the timedomain vibration signal singular spectrum entropy (TSE), power spectrum entropy (PSE),wavelet singular spectrum entropy (WSE), wavelet energy spectrum entropy (WEE), wavelettime entropy (WTE) feature extraction; and on the basis of analysis of the fuzzy C-meansclustering method in fault diagnosis application and excellent shortcomings, and make furtherimprovements, using the method of fuzzy clustering to C nuclear troubleshooting; and finallyby DS evidence and BP neural network to achieve the optimal integration of the system,reducing the uncertainty of a single diagnostic methods, improved fault diagnosis accuracy of the results, in order to achieve a complete diesel engine fault diagnosis method of diagnosis.