Forecasting Day Ahead Spot Electricity Prices Based on GASVM

Abstract

Price is the key index to evaluate the market competition efficiency and reflects the operation condition of electricity market for electricity market decision-making. This paper illustrates the characteristics and methods of the electricity price forecast. In this article, we forecast electricity spot prices at a daily frequency based on one new classification techniques: genetic algorithm improved least square support vector machines (LSSVM). As a benchmark, an artificial intelligence neural network is used as specification. We find that in forecasting of the electricity price, in general ANN is not good enough, but the improved nonlinear regression of LSSVM forecasts are more accurate than the corresponding individual forecasts. Based on the characteristics and contributing factors of electricity price, this paper introduce a better method for electricity price forecasting, Finally, key issues in the electricity price forecasting are discussed whilst some hot topics for further work are also presented.