Automatic condition monitoring of grid-connected PV system using signal processing techniques and machine learning algorithms

Abstract

In electrical energy production field, the early detection of Grid-connected PV system faults is crucial to avoid any failure in the system camponants, which can lead to unexpected breakdowns that causes high repair costs and enormous economic and commercial losses. During PPT modes operation the system faults remain undetectable for longer periods introducing many threats to the system. This work presents an approach for faults detection in (GPV) system under Maximum Power Point Tracking (MPPT) mode during large variations of environment conditions. We propose an intelligent method based on signal processing techniques and Machine Learning algorithms to detect and diagnose the systems faults using the extensive measurements obtained from a GPV system under Maximum PPT (MPPT). The recorded scenarios include seven faults: open circuit in PV array, grid anomaly, inverter fault, feedback sensor, MPPT controller and boost converter faults