Research And Implementation Of Power System Short Term Load Forecasting Method

Short Term Load Forecasting (STLF) is one of the most important contents of running and dispatching power system. It is a very important aspect of power system to ensure operating safely economically and achieve scientific management in the power system. And it is one part of energy management system as well as a necessary content of the electricity marketplace operation management.Many methods for short term load forecasting have been proposed, for example, time series method, artificial neural network method, expert system method and so on nowadays, but there isn't a method which can obtain the satisfied forecasting results on any occasion for the reason of lack of accuracy. How to improve the precision of forecasting is the hot problem studied by counterpart expert all the time.The result of load data statistics shows that when electric load are scaled, the load scaled curves between similar day types are similar, but load scaled curves between different day types are different, on basis of which short-term load forecasting can be divided into two steps which are the forecasting of load scaled curve and the forecasting of day maximal load and day minimal load. Based on the above analysis, this paper proposes one short-term load forecasting method with radial basis function network. According to similar relation between load on weekend and load on festivals or holidays, and temperature on festivals or holidays has great impact on the characteristic of load, a kind of practical load forecasting method on festivals or holidays, which considers load on weekend as the foundation load and revises it according to temperature, has been proposed.Outlier identification and load forecasting modeling are the two important steps in short-term load forecasting. There is no acknowledged effective method for outlier identification in electric load data. A new method is presented to identify outliers in load data by fully utilizing the features of electrical load curves. First, the day load curves are clustered by k -means algorithm. Then the outliers in the curves included in the corresponding cluster are identified with each typical load curve. At last, the outliers are adjusted with typical curves data. Test resultsusing actual data are served for demonstrating the feasibility of the proposed method.Taking account of the demand of electricity market in China, a design of short term load forecasting system is presented in this paper. This system consists of three modules: original data maintenance module, load forecasting module and forecasting result-processing module. The load forecasting module consists of many models and each model can be independently used or some of them can be combined for integrated forecasting. With real load data of Henan Province and the city of Pingdingshan, the test results indicate that satisfactory forecasting accuracy can be achieved by this system.