Observer based fault detection filters for three-phase inverters and statcoms

The introduction of power electronics has brought great benefits in terms of size, weight, performance, and feasibility to various electrical systems. However, with increasing penetration of power electronics circuits in safety- and mission-critical systems, the reliability of the power electronics is becoming an issue of increasing concern. To increase the overall reliability of the power electronic circuit, one key area is developing effective mechanisms for fault detection and isolation. While average models have been proposed for fault detection, these average models disregard dynamics at switching frequency, which are sometimes crucial in determining the exact health of power electronic circuit components. This thesis develops and experimentally demonstrates a class of model-based fault detection and isolation filters for three-phase AC-DC power electronics systems that is able to detect all possible component faults, offers fast detection, and has the ability to capture slow degradation in individual components. The structure of these filters is similar to that of a piecewise linear observer, and in the absence of faults, the filter error residual converges to zero. On the other hand, whenever a fault occurs, by appropriately choosing the filter gain, the filter error residual will exhibit certain geometric characteristics that allow the fault to be detected and, in certain cases, also isolated. While the theory developed is general, the analysis, simulations, and experimental demonstration are focused on systems implementing three-phase AC-DC converters used, for example, in drive applications and distributed static compensators