Time Series based Short term Load Forecasting using Prophet for Distribution system

One of the threats to reliable power supply from power systems is unpredictable load demand. Thus, precise load prediction is essential. Machine Learning and Data Analytics are recommended methods. A time series forecasting approach is used to predict future load demand levels with a high degree of confidence. The Prophet tool developed by Facebook is used in the process. The prediction model is developed using real-time hourly data from HESCOM for a stipulated time interval. The effect of various seasonality, climatic parameters like temperature, humidity, and precipitation is also considered. The model developed is assessed by the values of RMSE, MAE and R2.