Fault diagnosis via PV panel-integrated power electronics

This paper presents the design, analysis, and implementation of a fault diagnosis method for photovoltaic (PV) energy conversion systems. We present a model-based state estimation approach for detecting and identifying three types of faults — (1) converter input faults (e.g. faults in a PV panel), (2) converter component faults (e.g. switch faults or passive component degradation), and (3) sensor faults (e.g. voltage and current sensors) for PV panel-integrated power electronics systems. The state estimator model includes a dynamic model of the PV source and a linear-switched model of the switching power converter. The estimated state values are compared with measured values from the physical power stage, which generates an error residual vector. This residual is used to detect and identify faults in either the PV source or the switching power converter. The estimator, fault detection and identification logic, and the PV converter control system (including PWM generation and maximum power point tracking) are implemented entirely on a single all-programmable system-on-chip (SoC) device, which includes an FPGA and ARM core. We present simulation and experimental results for a prototype 2 kW PV energy conversion system to demonstrate the efficacy of the proposed fault diagnosis and control platform. The experimental results demonstrate successful fault detection within 2 ms and precise fault identification within 31 ms for a collection of input, component, and sensor faults.