Forecasting of Wind power using Variational Mode Decomposition-Adaptive Neuro Fuzzy Inference System

Abstract

According to Central Electricity Regulatory Commission, India, all independent power producers should forecast their generation and submit a report regarding the same to RLDC (Regional Load dispatch Centre). If a deviation occurs between forecasted and actual generated power, the renewable energy operators should give penalty to RLDC. In the wind farm scenario, the wind farm operator should predict the wind power accurately to reduce the risk of uncertainty and penalties. To estimate the wind power precisely, the wind farm operators will depend on commercial forecasting methods. The selection of forecasting method is based on forecasting accuracy, system availability, Lead time etc. The aim of this work is to do wind power forecasting using hybrid VMD- ANFIS (Variational Mode Decomposition-Adaptive Neuro Fuzzy Inference System) in different time horizons. The power data for two years is obtained for a site in Maharashtra having 15 wind turbines, each having a capacity of 800kW. Three evaluation indices such as Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE) and % Revenue loss are calculated for one hour ahead and one day ahead forecasting and results are presented.