Online Multi-Layer Perceptron-Based Power Quality Monitoring for Electric Vehicles

Electric vehicles can cause power quality degradation during the charging and discharging operations of the battery bank in grid-to-vehicle (G2V) and vehicle-to-grid (V2G) enabled vehicles. Sudden decreases in voltages, overvoltages, and grid interruptions (power quality disturbances or PQDs) have to be identified in order to prevent a major grid failure. This paper proposes a multi-fault classification/monitoring system in a power grid connected converter widely used in the charging/discharging system of electric vehicles; this is achieved by a machine learning classification algorithm called multi-layer perceptron. The main objective of this paper is to develop a neural network that monitors the signals coming from the power grid in real-time and can classify behaviours that may cause a malfunction during the operation. Therefore, this study can be used as a monitoring system that can, in a future, perform corrective or/and preventive actions, ensuring a regular operation when PQDs occur in the power grid.