Data-driven design and robust implementation of monitoring and fault detection system for AMT vehicles

Automated manual transmission (AMT) and its automatic shifting is of significance on the improvement of driveline performance and reliability. Due to numbers of electrical, mechanical and vehicle longitudinal dynamics, both establishing an analytical AMT model and designing its monitoring and fault detection (MFD) system become too complicated to implement for automotive engineers and servicemen. To solve the problem, in this paper, the data-driven design approach is introduced. The central idea behind the scheme is to construct a parity vector-based residual generator from test data and, based on on-line ones, to realize a robust adaptive residual generation against changes along with driving conditions. A simulation study using a driveline model constructed by the commercial software AMESim is given to illustrate the availability of the designed MFD system.