Abstract
In this study, the model-based fault detection and isolation (FDI) approach of parity-space is adapted to the diagnosis of sensor faults in power systems. Hardware redundancy is conventionally utilized to overcome this problem. However, this is an expensive solution. Instead, we propose to detect and locate faults by the systematic use of the system's analytical redundancy, with a global view of the system. This redundancy can be used to detect and isolate sensor failure as well. We also give necessary and sufficient conditions for a sensor network to be able detect faults, sensor failures included. Hence, the sensor configuration problem boils down to an optimization problem that can be intelligently guided by our method. The principle of parity-space approach is described in detail and illustrated on a simple power system model. The method is then validated through simulation.