Combined Forecasting For Medium And Long Term Load Based On Grey Model

Medium and long term load forecasting, as a fundamental item of the urban power system planning and a precondition of safe operation of power system, whose accuracy has great influence on the quality of power system planning, is a complicated project overloaded with details, especially having excessive uncertainty and involving extensive domain. How to design an effective and applicable medium and long term load forecasting software system for urban power networks attracts a lot of concern of electric operators. Nowadays, conventional methods, which have some deficiencies, for example, needing more historical data and having low precision of forecasting results, haven't been accepted gradually. In contrast, grey model (GM), which needs less historical data and has fine precision of parameters estimation, becomes more and more current. The algorithms of medium and long term load forecasting system for urban power networks are studied, a combined optimum model with high precision is put forward, and the forecasting software is programmed in this paper.Medium and long term load of power system has two layer trends of increscent and fluctuant at the same time. The deficiency of the basic grey model and other improved models are analyzed, and a new method which is called a second combined model based on the combined optimum grey model is also introduced in this paper. Two parts compose the combined grey model in this paper. One is combined optimum grey model including partial error grey model and equal dimension and new information grey model. Another is linear regression analysis method. The partial error grey model has a character of fitting wave character load. The equal dimension and new information grey model can remove outdated data, add the newest data and renew the data base in order to reduce computational complexity and improve the precision of forecasting model. And the linear regression analysis method can improve the precision of forecasting model by the method of considering all kings of factors that influence electrical load. The calculation results show that forecasting power load by combined model is credible and simple. The combined model can estimate the model parameters, meet the requirement of dynamic power load and solve the problem of great effect of random disturbance. For this type of complex problems, the combination grey model is especially useful because of its high precision and facility. The method can be used as one of the tools of forecasting the medium and long term power load.According to the practical demands of electric departments, whole medium and long term load forecasting software for urban power networks is developed. It has been proved by using practical data that this system can commendably satisfy demands of load forecasting for the urban power networks planning and increase planner's work efficiency, with friendly man-machine interfaces, convenient accesses and complete graphical functions. The error of the forecasting software can be controlled in the domain of±3%.