A novel hybrid method of LSSVM-GA with multiple stage optimization for electricity price forecasting

Abstract

Predicting price has now become an important task in the operation of electrical power system. Day-ahead prediction provides forecast prices for a day ahead that is useful for daily operation and decision-making. The main challenge for day ahead price forecasting is the accuracy and efficiency. Lower accuracy is produced due to the nature of electricity price that is highly volatile compared to load series. Hence, some researchers have developed complex procedures to produce accurate forecast while considering significant features and optimum parameters. Therefore, a multistage optimization for hybrid Least Square Support Vector Machine (LSSVM) and Genetic Algorithm (GA) model is developed in this study to provide an accurate price forecast with optimized parameters and input features. So far, no literature has been found on multistage feature and parameter selections using the methods of LSSVM-GA for day-ahead price prediction. All the models are examined on the Ontario power market; which is reported as among the most volatile market worldwide. A huge number of features are selected by two stages of optimization to avoid from missing any important features. The developed LSSVM-GA shows higher forecast accuracy with lower complexity than the existing models.