The Application And Research Of Self-learning Algorithm In The Clutch In AMT System

Compared with Hydraulic Automatic Transmission and Continuously Variable Transmission, Automatic Mechanical Transmission (AMT) has advantages of higher efficiency,easier installation and lower cost in manufacturing,so there is a vast potential market for AMT in this field. It is said that AMT will be the emphasis of automated transmission development in China. However, if the AMT want to develop, the first to be resolved is the control of the clutch engagement in vehicle start-up process, which is one of AMT technical difficulties. Because the determine of the clutch half-junction point and the bonding speed directly affect the quality of the vehicle start-up process in AMT system.Following items are included on the research of the clutch engagement control(1) According to characteristics and the bonding process of the clutch engagement, the various factors such as the spring diaphragm pressure, engine speed, and the opening degree of the throttle, etc., which need to be considered in the bonding process of the clutch engagement are analyzed. Then the difficulties of clutch engagement are determined.(2) Using the compensation method to improve the change of the clutch half-junction point which caused by the vehicle environment in the start-up process. The final verification by the real vehicle determine the feasibility of the compensation method.(3) The self-learning algorithm based on Human-Simulated Intelligent Control is used to control the clutch engagement for the engine runing at the constant speed in vehicle start-up porcess;Based on the analysis of the Human-Simulated Intelligent Control System such as characteristics,the feature model,feature identification,feature memory, the design process of the Human-Simulated Intelligent Control is given.Then, MATLAB simulation results the control effects.(4) The hardware and software of AMT system are realizeded for the compensation algorithm of the clutch half-junction point and the self-learning algorithm of the clutch engagement control which based on the Human-Simulated Intelligent Control under the condition of the engine running at the constant speed. The overrall structure of the system is introduced and the hardware and software are designed; Freescale's MC9S12C32 microcontroller is used as Electronic control unit.C programming language is used to design the electrical controlled software and the modular structure design method of the software can greatly simplify the difficulties of programming,improve the readablility, maintainability and portability,which ensure the scalability and reorganization of the software.According to the actual experience,the requirements of the reliability of hardware and software is given.