Fault Monitoring of the Electric Machine in a Hybrid Vehicle

A diagnosis system for the electric machine and the power electronics in a hybrid electric vehicle is designed, where the diagnosis system uses a map based model of the system to be monitored. Such a model gives an accurate description of the fault free system, and is therefore suited for fault detectability. However, one drawback using such a model for diagnosis is that fault isolation often requires that the model, in addition to the fault free case, also describes the faulty system, and thereby measurements of the faulty system are needed, which is costly. Another approach is to use a model including physical parameters of interest in the system to be monitored, to also describe the faults’ impact on the system. To achieve good diagnostic performance such a model needs to be accurate, which also is costly. Therefore, in the new approach taken here, two models for the system are used in combination to achieve good fault detectability and isolability; one is a map based model, and one is describing the faults of the system. It is shown that the approach works well and is a promising path to achieve both good fault detectability and isolability performance, without the need for neither measurements of a faulty system nor detailed physical modeling. In a simulation study evaluating the designed diagnosis system all faults are isolated and also accurately estimated.