Optimal combined load forecast based on the improved analytic hierarchy process

• Main Authors: Zhou Ren-jun, Duan Xian-zhong
• Format: Conference Proceeding
• Language: English
• Subjects: Economic forecasting; Educational institutions; History; Load forecasting; Mathematical programming; Matrix converters; Neural networks; Power engineering and energy; Power system analysis computing; Power system planning
• DOI: 10.1109/ICPST.2002.1047570

Different forecasting methods can lead to very different results in power system load forecast. In order to improve forecasting accuracy, combined forecasting methods have been introduced in solving the engineering forecasting problems. In this paper, the conception of soft method is first presented in power system load forecast. The computing and analyzing methods in soft science such as decision-making analyses have been introduced to solve load forecast that are unstructured problems of multi-factors. The combined forecasting problem is treated as multi-hierarchies and multi-factors evaluation by composing qualitative analyses and quantitative calculation. In addition, the experiences and judgments of experts will be collected to implement judgment matrices in group decisionmaking. The soft method based on improved analytic hierarchy process (AHP) is proposed to carry out long-middle term load combined forecast in this paper. A hierarchy structure has been established by analyzing various factors that affect the load forecast. It is the key to determining the combined weight coefficients in the optimal combined forecasting method. Fuzzy complementary judgment matrixes of pairwise comparison will be formed by the expert in each hierarchy and be converted to a fuzzy consistent matrix. The eigenvector can be calculated using its general formula and be regarded as weight coefficient in combined forecasting. The combined forecast methods based on the improved AHP is of clear hierarchy structure, sufficient judgment information and simple calculation formula. The forecasting examples show that this method is practical, convenient and accurate.