Enhanced parity equations method for sensor fault detection in electrical drives

This paper focuses on sensor fault detection and isolation (FDI) problem by enhancement of a parity space approach. Indeed, faulty sensor can degrade overall system performances, especially in electrical drives where inner current control loops are very fast and based on sensors outputs. Parity space theory is investigated in the context of general FDI algorithm generation for second-order systems, which are typically used for electrical drives analysis, by starting from higher-order models. The study focuses on simplifications introduced by real-time implementation considerations. A simple and fast algorithm is derived when sampling time is very low, which allows the integration of the FDI algorithm between two consecutive control sampling times. Moreover, the proposed algorithm is insensitive to motor parameters variations and available for any kind of sensor giving an output electrical signal. The development is illustrated with the case of a DC drive. Simulation results are given to demonstrate the effectiveness of the proposed approach.