A study on self-learning control strategy for AMT clutch key control parameter

Clutch kiss-point has static deviation caused by manufacture error and dynamic deviation caused by friction plate wear. Without corresponding sensor, input shaft speed is calculated by vehicle speed. Maximum crankshaft acceleration and hysteretic nature of input shaft speed are analyzed. To overcome static deviation, an offline self-learning strategy is proposed, in which clutch engages under fixed-step to overcome the above-mentioned hysteretic nature. To overcome dynamic deviation, an online adaptation algorithm is proposed, in which kiss-point is identified by input shaft speed delay time and maximum crankshaft acceleration. Verification tests on vehicle show that the strategies achieve expected effects.