Research on Intelligent Self-Learning System for the Electro-Pneumatic Automated Manual Transmission Gear Position

Shaohua Sun,Shunbo Li,Yao Fu,Cheng Yang

DOI: https://doi.org/10.4271/2014-01-1741

2014-01-01

Abstract:In the electro-pneumatic automated mechanical transmission (AMT) system, the manufacturing, assembly, wear, replacement and other issues often lead to gear position change and some differences in various gears of transmission, which reduces the success rate and even results in abnormally working. To solve these problems, based on the intelligent control theories of fuzzy PI control, this research developed an intelligent self-learning system for the AMT gear position. This system contains the position initialization module after assembling (offline module) and the position correction module in use (online module). The system can automatically recognize gear position deviation and actively correct it, which improves the robustness of shift actuator. The precise control of shift actuator is the key of this self-learning system. However, due to the compressibility of air and uncertainty of external flexible dynamics, it is difficult to establish accurate models of the system in each stage, so conventional PID control methods have been unable to accurately control the position of shift cylinder. In this study, the flow equation of shift valve is established based on analysis of the operating principles of electronically controlled pneumatic AMT, and then a fuzzy PI control strategy is developed according to fuzzy control theory. With the platform of pure electric car equipped electrically controlled pneumatic AMT, a series of field vehicle tests were carried out and the results showed that the development of intelligent AMT shift position self-learning system can effectively achieve self-learning of shift position accurately with good consistency and higher efficiency.