Energy Management Systems For Short-term Load Forecasting Based On Support Vector Machines

Anshan Iron and Steel Group Corporation, one of the largest and prestigious international mining companies, whose annual turnover is more than 150 billion Yuan, is energy-hungry and also has a potential ability in saving energy. In order to improve management of enterprise and the efficiency of energy, Anshan Iron and Steel Group Corporation cooperated with Changchun University of Technology to develop the Energy Management Information System of Dagushan concentrator. Short-term load forecasting is an important component of energy management system, in this paper, we focused on the he short-term load forecasting of energy management information system. By analyzing and predicting the power, the system could offer the cost of power of Anshan Iron and Steel Group Corporation, direct the running program and the supplying power plan, encourage enterprises to use energy rationally and reduce the cost of energy using, and improve the competitiveness of enterprises.Short-term load forecasting refers as the load forecasting one day to several months in the future. The load curve forecasting of the next 24 is the most important of it which is also the part this paper analyzed.Although the load forecasting research has been for several decades of history, new theory and new technology based load forecasting researches have been developed continuously. As new technology of data mining, support vector machines (SVM) have been successfully applied in pattern recognition and regression problem and so on. This paper analyzed the basic principles of support vector machines electricity and load forecasting for concentrator power load characteristics, focused on analyzing its main influencing factors. By comparing the different kernel function, parameters of the forecast results, the paper established the optimal kernel function, related parameters and accomplished a short-term load forecasting model SVM. We also analyzed and compared traditional BP neural network with this improved method by experiment, the results show that the accuracy and speed of support vector machine-based load forecasting are superior to that of neural network method.Because of numerous load influence factors having a great of complex characteristics, the pattern, without selecting input vectors, will lead to reduce of the precision and increase of the compute running time. Therefore this paper adopted an effective fuzzy clustering analysis and process technology, proposed a new method for short-term load forecasting which combined fuzzy C means clustering algorithm (FCM) with support vector machine. This method chooses training samples by fuzzy clustering according to similarity degree of the input samples in consideration of the periodic characteristic of load change, which means take the same type of the data as the learning samples for forecasting, guarantee the consistency of the data characteristic and enhance the history data regulation. The results of the practical application of the proposed method show the usefulness of this method, both the precision and speed of load forecasting can be improved. This method is fully applicable to the Anshan Iron and Steel Mining Company's forecasting and analyzing of power consumption...