Identification of fuzzy model for short-term load forecasting using evolutionary programming and orthogonal least squares

This paper presents a weather sensitive short-term load forecasting (STLF) algorithm, based on a novel fuzzy modeling strategy using evolutionary programming (EP) and orthogonal least squares (OLS). Traditional forecasting models based short-term load forecasting techniques have limitations especially when weather changes are seasonal. The proposed fuzzy modeling strategy mainly contributes to predicting the hourly load when the load change is influenced greatly by temperature. In this paper, the OLS method is applied to select the input terms for the consequent part of the fuzzy rule base evolved by EP without changing the premise part. The parameters identification to the consequent part is completed simultaneously by the OLS method. Observing that the fluctuation degree of the temperature load curve is much lower than that of the load curve when temperature greatly influences the load change, we utilize the relative load variables as one part of consequent input candidate set to STLF fuzzy model. This method was tested on the practical load data of Zhejiang Electric Power Company in China. The testing results demonstrate the great contribution of these relative load variables to better forecasting performance. And the superiority of the proposed method is also demonstrated especially when the load change is greatly influenced by the weather terms