Application of maximum entropy principle to load forecasting collaboration

Abstract

This paper proposes a collaborative load forecasting method based on maximum entropy principle. Taking the expectation and second order central moment of forecasting results into account and using the maximum entropy principle, the load forecasting method receives the probability distribution function of the forecasted values and obtains the final high, middle and low forecasting scheme based on the probability theory. Based on this method, three applications in load forecasting are introduced in this paper. Case studies have shown applications of this method can take the affection of the conflicts among different forecasting path into account in forecasting, resulting in higher precision of load forecasting.