Research And Application Of Quality Prediction And Diagnosis Method For Gearbox Housing Machining Process

As an important part of the vehicle,the quality of the dual-clutch transmission(DCT)directly affects the overall performance of the vehicle.And as a key component on the DCT,the transmission housing will also affect the assembly quality of the transmission.Therefore,this dissertation establishes a quality prediction and diagnosis model based on the Factory Automation System for the key station of the dual-clutch automatic transmission from the casing machining process.In this dissertation,the key work center in the casing machining process is taken as the research object.Firstly,the dissertation introduces the machining process of the casing and has an analysis and introduce on the requirements,modules and key technologies of the casing machining quality control system.Secondly,for the processing process,this dissertation proposes a prediction model of the divulging test results of the gearbox shell based on the dynamic learning rate BP neural network algorithm.And the actual processing data is compared with the basic BP neural network algorithm in predicting accuracy and convergence speed.According to the comparing,the validity and improvement of the dynamic learning rate BP neural network quality prediction model are verified.For the multivariate quality diagnosis in the machining process,this dissertation,then,constructs a multivariate residual control chart diagnosis algorithm based on multi-stage least squares support vector machine(MS-LSSVM).Through the classification and learning of historical data,this dissertation establishes a multi-stage quality diagnosis model based on different data characteristics and achieves the purpose of diagnosing abnormal variables according to different data categories,guiding relevant personnel to carry out process improvement,and ultimately improving product quality.Finally,the actual data of a company is taken as an example to verify and compare with the least squares support vector machine algorithm model.It is verified that the algorithm model based on data classification has better performance in diagnostic accuracy.At the end of the dissertation,a factory automation system was developed and applied to an automobile gearbox factory,It provides data foundation and technical support for the factory to promote digitalization,intelligence.