A method of optimal shift control based on pattern recognition and learning algorithm

Abstract

Aiming at the problem of optimal shift control of the automated mechanical transmission (AMT) used in an automobile, a method of optimal shift control based on pattern recognition and a learning algorithm is presented. A pattern recognition technique is applied to simulate the intuitional thought of a driver in the process of deciding to change the gear position of a manual transmission, and an iterative learning algorithm is employed to make the shift schedule approach the optimal schedule in the process of repeated shifting, so that an optimal shift schedule need not be established in advance. Determination of the optimal shift schedule depending on simulation tests with transmission system test equipment can be avoided and manpower spared, expenses reduced and the time of developing AMT products shortened.