Research And Engineering Application Of Fault Diagnosis Method For Piston Engine

As a common piston engine, diesel engine is extremely extensive in the production of industrial, mainly used in petroleum, electric power, metallurgy and other fields. Therefore, to ensure its safe, stable and efficient operation, monitoring the status of the diesel engine has a very important safety and economic value.The fault diagnosis method based on the vibration signal processing of diesel engine has became one of the commonly used methods of fault diagnosis, for the information of the vibration signal of diesel engine is abundant. Because the diesel engine cylinder head vibration signals contains large amount of state information, the vibration signal processing technology based on diesel engine cylinder head has became the preferred method.Compared to the fault feature extraction method, the intelligent diagnosis method does not need to extract the fault feature, which can judge the running state directly according to the state characteristic information of the mechanical equipment.Commonly used intelligent diagnosis algorithms are:neural network, clustering analysis, support vector machines and so on. The fault pattern recognition method based on cluster analysis has a serious impact on the result of fault diagnosis due to the randomness of the initial value selection. When the number of samples is large, the efficiency of fault diagnosis is seriously decreased, and it is difficult to meet the real-time requirements of fault diagnosis in real engineering. The fault condition recognition method based on neural network has the complexity of the algorithm and the large amount of computation, which leads to the recognition efficiency of the method is low, and the recognition result is easily affected by the number of training samples. Compared with the previous two intelligent identification methods, SVM is more suitable for processing small samples, which overcomes the problem that fault samples are insufficient in engineering practice.Diesel engine is an important source of power of many large equipment, due to the backward practical means of monitoring and fault diagnosis, engineering application in many malignant frequent breakdown, the safe and stable operation of equipment is important have considerable influence. Study of diesel engine fault diagnosis, the experimental diesel engine fault early warning and fault diagnosis has very important theoretical and practical value. Fault simulation experiment is an important part of the research of fault diagnosis, research on fault mechanism and signal characteristic provides a large number of actual case data.A new fault diagnosis method based on principal component analysis (PCA) and support vector machine (SVM) was proposed in this paper. First, a number of time domain parameters of the vibration signal were extracted, and the frequency domain features were extracted by wavelet packet decomposition, then used PCA to extract the time domain and frequency domain feature to select the sensitive feature, reduced dimension processing and the complexity of data processing, finally, SVM was used to train and test the feature subset. The diagnostic accuracy of this method in typical practical faults of diesel engine was as high as 98%, the validity of the method is confirmed.