An Efficient Short-Term Electric Power Load Forecasting Using Hybrid Techniques

Abstract

The power system plays an essential part in the transmission of electricity. Generation, transmission and distribution are the main parts of electricity consumption, and day to day, the demand for electricity increases. So, the prediction of price and load is significant in the power system. The short term load forecasting (STLF) predicts the load 24 hours ahead or a week ahead. This paper investigates the effect of price and load on short term load forecasting with two hybrid techniques (HT). The first hybrid technique consists of wavelet as well as Generalized Autoregressive Conditionally Heteroscedastic (GARCH) analysis methods and the second hybrid technique consists of GARCH, EGARCH, GJR models and Particle swarm optimization (PSO). The first hybrid technique (FHT) is analyzed by taking the Price from the Ontario grid and the second hybrid technique (SHT) is diagnosed by taking the load from the Xintai power plant. The prediction of Price and load depends on input data. The results are significant and its accuracies are effective. The analysis of two hybrid techniques clearly defines its work and the ability to produce the result. In the first HT, the wavelet decomposes the data and the GARCH models analyze the decomposition data. In the second HT, the GARCH, EGARCH, GJR models are executed the input data and PSO brings the smoothness of the result. The forecasting price results and load calculation indicates less error and it is essential for short term load forecasting.