Research On Self-learning System For Clutch Pressure Control Of Automatic Transmission

With the continuous development of the automobile industry,the automobile equipped with automatic transmission has become the mainstream.Among them,automatic transmission(Automatic Transmission,AT)has become the first choice for off-road vehicles and commercial vehicles due to its good shifting smoothness,stable start and excellent acceleration performance.The key technology of AT lies in the reasonable selection of shifting points and the effective control of shifting process.The control of the shifting process directly affects the shifting smoothness of the transmission,the service life of the components of the transmission,and the riding experience of the driver.This paper relies on an Army equipment project "New Type ********** Automatic Transmission(AT)Integrated Application Technology”.And the purpose is to improve the shifting process,reduce the impact and shift time;maintain the car transmission with good shift quality throughout the life cycle.Based on the research of objective shift quality evaluation of shifting process,an AT clutch(brake)pressure control parameter self-learning system is established.The main research contents are as follows:The dynamic characteristics of AT transmission system including planetary gear,hydraulic torque converter and clutch are analyzed and simulated.Considering the power input of the engine and the load of the vehicle,the vehicle model is established.According to the operation principle of solenoid valve,sliding valve and clutch in electro-hydraulic control system,an accurate model of electro-hydraulic control system is established.Using MATLAB and AMESim software,the vehicle model and the electro-hydraulic control system model are simulated jointly to verify the validity of the model.This paper introduces the traditional evaluation index of shift quality of automatic transmission,and puts forward the specific evaluation index of AT according to the power transmission characteristics of AT.The control methods of clutch-to-clutch shift process including torque phase,inertia phase and end of shifting process are introduced,and the influence of error of clutch torque control on shift process is analyzed.According to the characteristics of the electro-hydraulic control system,its structural parameters is analyzed include friction disc clearance and main oil pressure,as well as control parameters including oil filling time,holding pressure,clutch transferable torque and closed-loop control parameters,and the influence of changes in different stages of the shift process on the clutch torque is analyzed.The AT shift control system is established based on vehicle model and electro-hydraulic control system model.Based on the model of AT shift control system,Kalman filter algorithm is used to process the speed information of the input and output axes of AT,and real-time analysis of the state of AT is carried out.According to the real-time state of AT,the off-line self-learning is carried out,and the relevant control parameters of the electro-hydraulic control system are obtained,which can be applied to on-line control.A clutch pressure self-learning algorithm for AT shift process is proposed based on Human Simulated Intelligent Control Algorithm and Iterative Learning Algorithm.The algorithm can automatically identify the problems in the shift process according to the evaluation index of shift quality and the data processed by Kalman filter,and select the corresponding control parameters for online self-learning,so as to achieve the optimal control effect,so that the system has good shift quality.The simulation results show that the clutch pressure self-learning control algorithm can effectively control the model when the control parameters and model parameters change,and has a good control effect.Based on TCU hardware platform and CANape calibration system,the effectiveness of clutch pressure self-learning control algorithm is verified on vehicle test platform when control parameters change.According to the actual monitoring data,the shift process is evaluated.The experiment shows that the algorithm can effectively improve the shift quality.