Research Of A Medium And Long-term Load Forecasting Method

Medium and long-term load forecasting of power system is defined to be the forecasting of power load, quantity of electricity and other stuff in more than one year. As the long-range plan of power system, medium and long-term load forecasting provides elementary data for power network planning, and at the same time it is the very important basis for the arrangement annual maintenance plan and the way to run. Network planning department provides the results of medium and long-term load forecasting based on situation of national economic development, so that the power network is able to be renovated and expensed. The main factor which affects load growth includes two categories. The first one is easy-quantified factors, which includes economic development, population, consumption per unit output value, industrial restructuring and etc, and the other one is hard-quantified factors, which includes political situation, weather conditions, pricing policy and etc. It has been much harder to predict precisely because of the high marketing level of power system.Based on generalization of the course of development , the accomplishment in the past few decades and the development trend of prediction methods on this thesis, the advantages and disadvantages of various mature methods and its scope of application is relatively compositely compared and synthesized. In this paper cointegration test and correction mode which have been widely used in economic and import-export forecasting are applied here, and their applicability is analyzed. At the same time genetic algorithm, co integration test, error correction model are combined to establish a new medium and long-term load forecasting model. The new model established is used to forecast the load of a province in China and its prediction accuracy and adaptability are evaluated and estimated.