Fault Diagnosis of Diesel Engine Based on EMD Decomposition
|School||Jiangsu University of Science and Technology|
|Keywords||Diesel engine Fault Diagnosis EMD Correlation dimension AR model Neural Networks Support Vector Machine|
Diesel engine as a complex power machinery , widely used in ships, locomotives , cars and generator sets running state directly affects the security and reliability of the power system as a whole . The diesel engine will definitely vibrate reflect its performance of the work of the internal parts of the state information through certain pathways to the surface of the vibration signal , it can be used in diesel engine vibration signal disintegration fault diagnosis . Diesel engine fault diagnosis technology , made ??a more in-depth discussion of the cylinder head vibration signal based fault feature extraction and its diagnosis . The main content of this paper includes the following aspects : first , both from the theoretical and experimental analysis of the characteristics of the internal combustion engine cylinder head vibration signal reveals the characteristics of the vibration source , the diesel engine failure mechanism of the route of transmission , and to discuss the time domain and frequency domain characteristics of the diesel engine cylinder head vibration signal . Second, EMD decomposition of the cylinder head vibration signal with EMD , explore the decomposition of the physical meaning of each IMF component , try the IMF component HHT marginal spectrum analysis to find fault feature . Third , the use of time series analysis methods and fractal theory analysis of the vibration signal feature strike order IMF component of the AR model parameters and associated dimension , to verify the different conditions the intrinsic relationship between the amount of the state with the characteristics , which separating the independent features to reflect the failure of the feature information and fault information . Fourth , the substance of the conditions state recognition status classification problems , the AR model parameters and correlation dimension as the feature vectors imported into the neural network and support vector machine training , to identify the state of the diesel engine fault condition to verify the neural network the feasibility of the application in diesel engine fault diagnosis and support vector machine , in contrast to higher recognition rate of the support vector machine is more suitable for small sample analysis . By theoretical calculation and analysis , the diesel engine valve gap abnormal as well as the failure of the oil off vibration diagnostic mechanisms and diagnostic methods , certain conclusions and methods of engineering value , this important reference significance for diesel engine fault diagnosis .