Study On Fault Diagnosis Technology Of Small Piston Engine Based On Optimized SVM

The working state of the piston engine directly affected the safe operation of the equipment.If the engine fault was not found in time,it could lead to the failure of the equipment to start or run normally,or the serious accident of overall damage to the equipment in the process of working.Therefore,it was of great significance for the safe operation of the engine to find the potential faults and maintained them in a timely and accurate manner.A fault diagnosis technique based on optimal SVM was studied for a small two-stroke piston engine.Firstly,according to the fault types in the actual use of the engine,the common oil supply system fault was selected as the research object for the data collection of the engine fault test.In view of the fact that the actual engine fault diagnosis was a multi-classification problem,the nonlinear support vector machine based on DDAG SVM was selected as the classification model,and the radial basis(RBF)kernel function was used to solve the problem of low dimensional inseparability of the data set.Secondly,the engine data acquisition system based on Lab VIEW was designed.The selection of various sensors in the engine data acquisition system was carried out,the signal conditioning circuit of the encoder was designed,and the transmission equipment matching the encoder and the engine was designed according to the engine structure,and the upper computer software with high-speed acquisition,processing and storage function was designed by using Lab VIEW.The test verified that the engine data acquisition system meets the requirements of engine data acquisition test.Then,the engine oil supply system fault was caused by fault injection,the test data was obtained,and the thermal parameters recorded in the test were analyzed.The in-cylinder pressure and the incylinder pressure rising rate under normal conditions were compared and analyzed from the perspective of time domain.The VMD variational mode decomposition method was used to separate the characteristic signal from the noise signal from the engine fault original data in time-frequency domain,and the corresponding relationship between the engine cylinder pressure and cylinder head vibration data was identified.Finally,using the singular value and energy feature extraction method to extract the feature data and feature data set was established,the normalization processing of data set was divided into training set and test set,fault diagnosis of the SVM model was optimized by using,and then through the cuckoo search algorithm combined with the training set to optimize the SVM model,through the comparison between results of fault classification,verified the CS-SVM optimization effect.Of engine fuel supply system of 8 different work condition in cylinder pressure data and comparing for fault diagnosis of engine cylinder head vibration data,verified the engine cylinder head vibration signal of the effectiveness of the small piston engine fuel supply system fault diagnosis,but also illustrated the research of fault diagnosis method based on the optimized SVM has stronger practical value.