A GRU-GA Hybrid Model Based Technique for Short Term Electrical Load Forecasting

Load forecasting is an integral part of power system planning and operation. Precise forecasting of load is essential for unit commitment, capacity planning, network augmentation and demand side management. Amongst different forecasting techniques, hybrid models show better performance. In this report, an integrated model of Genetic Algorithm (GA) and Gated Recurrent Unit (GRU) network is proposed based on day lag feature for day ahead short term load forecasting (STLF). In conventional GRU network, the best window size, number of neuron, and other factors are usually selected by trial and error as well as researcher's experience. This study proposes an important approach for selecting the time lags and neuron number using GA. The hybrid approach is experimented on electricity load data of Bangladesh to evaluate the effectiveness of the technique. The outcome provides a significant improvement compared to a previous work on the same load data with the least MAPE and RMSE value of 0.83 and 68.46 respectively.