Research On The Engine Crankshaft Multistep Assembly Quality Prediction Model Based On LS-SVM Of Particle Swarm Optimization

Along with the rapid development of Chinese automobile industry, the complexity of the automobile engine as power generating plant is higher and higher, the corresponding difficulty of assembly technology is becoming higher and higher. The real-time online quality prediction is the key to the engine crankshaft multistep assembly quality control. The research method for prediction of quality is from expert system method to neural network method, and then to all kinds of support vector machine (SVM) method; For the perspective of the research, it is from the process level to the system level and from single process to multiple process. There is more and more quality prediction research. But for now, there are few studies on engine application with the quality prediction.For the engine crankshaft torque detection in multistep assembly error volatile problems, the engine crankshaft multistep assembly quality prediction model based on least squares support vector machines (LS-SVM) of particle swarm optimization (PSO) was constructed. Considered of the uncertainty of assembly quality and the relativity of assembly process characteristics, we selected the axial clearance, alignment, clearance fit, the bending quality of such main factors as input features, crank torque as the output quality. With the data for model training and learning from the field, we used particle swarm optimization algorithm of least squares support vector machine to optimize.Then we applied the trained model to predict the corresponding conditions of crankshaft torsional moment. In the end, taking the engine crankshaft torque testing companies in Chongqing as an example and analyzing the neural network model at the same time, the results show the practicability and validity of the model, and provide the reality basis from the Angle of the multistep for on-line quality prediction and control, so as improve the stability and efficiency of the automobile assembly process quality.