Ensemble approach for short term load forecasting in wind energy system using hybrid algorithm

Abstract

The uncertainty problem in the resources is essential to mitigate and to improve the system operation in order to attain the load forecasting. Sometimes, the wind power saturation level is high on the grid side; therefore wind power prediction is essential to improve the efficiency, safety, economic and stable operation of the electrical grids. In the wind energy system, balancing supply and load demands is being considered as a challenging task and it can be compensated by means of wind power forecasting. This paper depicts an ensemble approach for short term load forecasting (STLF) by means of the hybrid algorithm in the wind energy system. The hybrid algorithm is a grouping of the Deep Neural Network (DNN) and Chicken Swarm Optimization (CSO). Initially, 24 h load data of the wind energy system is collected from the New England ISO for training the DNN network thereby analyzing the load forecasting. During the training period, the training error rate is minimized with the help of the CSO algorithm. After the training period, there arises a testing period that recognizes future loads by means of the proposed hybrid algorithm. Based on the above consideration, the load forecasting problem in the wind energy system is achieved. The efficacy of the proposed method is expressed by computing the statistical measures in terms of Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE), respectively. The proposed method will be implemented in the Matlab/Simulink and its performances were compared with the existing methods such as DNN and ANN, respectively.