A Data Driven Method for Arc Fault Detection in Actual Photovoltaic System

In the actual outdoor photovoltaic system, it is hard to discover the series arc fault due to the weak variation of current, and is easily affected by the environment (light, temperature, etc.). To deal with this problem, the arc fault experimental platform with actual photovoltaic array is installed to collect the arc fault current. Then, the data driven mode is selected, and the temporal convolutional network(TCN) is used to build a fault detection model. The usability of TCN model is confirmed from the prediction outputs.