The Fault Diagnosis Of Diesel Engine Based On Fractal Technology And Probabilistic Neural Network

With the fast development of economic society, rapid development of manufacturingindustry, the fault diagnosis technology of Diesel engine developing rapidly, effectively notonly reduce the frequency of equipment failures, but also improve economic efficiency. Inthis paper, based on the previous research, the needs and requirements of consolidated thesubject, studing the characteristics of the diesel engine fault. Mainly to obtain the vibrationsignal from the signal source to analysis characteristics of the engine failure by the new ideaof fractal technology and the probabilistic neural network (PNN).In this paper, the diesel engine as the research object is studied. Firstly, the significanceand current situation of the machinery fault diagnosis are introduced, as the fractal theory andneural network applications significance. Being described the significance of wavelet theory,comparing wavelet noise reduction with wavelet packet noise reduction, to study the methodof the noise reduction to obtain the characteristics of the signal using wavelet packet, aboveall, it is better to access to the desired the effect signal by wavelet packet.Secondly, the paper introduce the measurement methods of the fractal theory, whichmethod is proposed will analysis the nonlinear signal well. Focus on the correlationdimension, the improved GP correlation algorithm is adopted, studied the delay time, thechoice of scale-free interval methods. As a visual representation of the correlation dimensioncharacteristic values can well reflect the characteristics of the nonlinear signal by extractingthe correlation dimension to identify the fault feature.Finally, in the part of this paper in detail introduce the characteristics of BP neuralnetwork and probabilistic neural networks (PNN), discovering the deficiencies of BP neuralnetwork, the method of the probabilistic neural network of fault diagnosis is proposed fortraining to accurately classify the fault samples. Illustrates the neural network can be used asan effective measurement of fault identification, which is very easy to classify the faultcharacteristics by the establishment of the training sample library. In order to determine the fault classification, the method of the fault diagnosis of diesel engine based on fractaltechnology and probabilistic neural network is used, which is verified feasible by analyzingand judging the characteristics of data.