Research On Fault Diagnosis Of Diesel Engine Based On CEEMD Sample Entropy

Diesel engine is a common complex power machinery, it has the characteristics of high efficiency and high specific power, so it is widely used in automobile, aircraft, ships and other transport vehicles. Whether the whole power system is safe and reliable, it is influenced by many factors, and the working condition of diesel engine is one of them. Thus it can be seen, it has great practical value that improve the diesel engine monitoring and fault diagnosis technology through the research. When diesel engine work, the status information of the internal parts will go through some channels to be reflected in the cylinder head vibration, so it is a kind of effective method to have fault diagnosis for diesel engine through the cylinder head vibration signal of diesel engine. The research on this topic mainly includes how to extract fault feature information from the diesel engine cylinder head vibration signals effectively and diagnosis of diesel engine fault state, A new method of fault diagnosis for diesel engine based on CEEMD- sample entropy is proposed. The work carried out in this paper is:(1)The experimental platform that can collect the information of the diesel engine cylinder head vibration is designed and constructed. The CZ4110 diesel engine is selected as an example to start the test, the data of the cylinder head vibration type diesel engine in different working conditions is collected successfully(including normal and abnormal state).It provides important for the cylinder head vibration signal feature extraction and fault diagnosis.(2)The reasons for the failure of diesel engine and transmission channels is researched.The characteristics of cylinder head vibration signal of diesel engine under different fault is analysised by combining theoretical analysis with experimental verification. By researching in the time domain and frequency domain, the characteristics of the vibration signal of the diesel engine is revealed in the two aspects.(3)The application of empirical mode decomposition EMD in the field of signal decomposition is studied. Taking into account that it has the problem of mode mixing in the process of signal decomposition in EMD, therefore, the EEMD and CEEMD decomposition method with the help of the noise is used. The two can suppress the mode mixing in a certain extent through experimental verification,so the method is effective,and experiments prove that CEEMD can decompose the signal into different time scales and extract the localinformation of the signal. The new method of CEEMD combining with wavelet is summarized to reduct noise, first, the signal is decomposed by CEEMD, then the noise reduction of each IMF is decomposed by wavelet, last the IMFs have been noise reducted is restructured as the final noise reduction signal. Experiments verify that the proposed method can effectively reduce the noise.(4)The sample entropy is used to measure the complexity and the nonlinearity of the signal. They are used in the measurement of the complexity of the vibration sequence of the diesel engine and the analysis shows that the sample entropy is consistent and affected by parameters and so on. Some points are put forward that the IMF components which are selected should be focused on the the intensity of the correlation between IMFs decomposed and the original signal. The cylinder head vibration signal in different frequency information can be obtained by using sample entropy to quantify the IMFs decomposed by CEEMD.They are put as the input vector of pattern recognition, it can provide the basis for diesel engine fault diagnosis.(5)The IMF sample entropy is used as the feature vector of the CEEMD decomposition,they are put into support vector machine for training and identificate the fault samples of diesel engine. By comparing with other diagnostic methods, it can improve the accuracy rate.The application of principal component analysis to the PCA principle in fault feature reduction is studied, Through the further study of the diagnosis experiment, it is proved that the method can effectively preserve the fault feature information and remove the redundant components, more accurate diagnostic information can be obtained. By combining with CEEMD-sample entropy method, good recognition result can be obtained in the fault diagnosis of diesel engine.