A parallel tabu search based fuzzy inference method for short-term load forecasting

Abstract

In this paper, a fuzzy inference method is proposed for short-term load forecasting. A new technique of parallel tabu search is used to deal with one-day ahead prediction of daily maximum loads. This paper focuses on a fuzzy inference approach due to good understanding of the nonlinear behavior of the model. Fuzzy rules help power system operators to explain their experiences and rules in an intuitive sense. In this paper, parallel tabu search is used to globally optimize the number and location of the fuzzy membership functions. It considers two strategies of the neighborhood decomposition and multiple tabu lengths so that computational efficiency and solution accuracy are improved. The proposed method makes use of the simplified fuzzy inference to alleviate computational effort for calculating the fuzzy membership functions of the output variables. The effectiveness of the proposed method is demonstrated with real data of Chubu Electric Power Company.