Diagnosis System Design Using CMAC-Based Scheme for Automobile Automatic Transmission

摘要

In this paper, a CMAC (cerebellar model articulation controller) neural network application on fault diagnosis of automobile automatic transmission system is proposed. Firstly, we build a CMAC neural network based diagnosis system with different coding scheme depending on the fault types. Secondly, the fault patterns, obtained from the China scholar's technical data, would be used to train the CMAC neural network off-line using new coding scheme. Thirdly, the learning algorithm was developed to guarantee the learning convergence. Finally, combining the MATLAB program the trained neural network can be used to diagnose the possible fault. Comparing with the traditional schemes, lower weights interference between different fault type patterns, higher noise rejection ability, do not require expert's expertise, fewer memory size and fast training speed are obtained.