Fault diagnosis in grid-connected PV NPC inverters by a model-based and data processing combined approach

Abstract

This paper presents a fault detection and isolation (FDI) method for open-circuit faults (OCF) in the switching devices of a grid-connected neutral-point-clamped (NPC) inverter for photovoltaic (PV) applications. The proposed methodology addresses the fault diagnosis problem by a combined model-based and data processing perspectives in order to study single and simultaneous faults in the NPC inverter. For the model-based scheme, a bank of sliding-mode proportional-integral observers is suggested to estimate the fault profiles under an additive model. Thus, from the estimated fault profiles, and by performing a directional residual evaluation in a fixed reference frame, single and simultaneous fault scenarios can be isolated in the NPC inverter. However, for some fault classes, there is some ambiguity by just the model-based approach that is overcome by employing the average line currents to construct extra fault signatures. The proposed FDI scheme only requires the measurements of line currents and grid voltages in the diagnosis media, and can isolate 6 × 2 single OCFs and 12 × 4 simultaneous OCFs in the order or lower than a fundamental period of the grid frequency. Our new FDI methodology is validated through experimental data from a practical PV system in a closed-loop grid-connected NPC inverter under single and simultaneous OCF conditions.