Application Of Support Vector Machines In Power System Short-Term Load Forecasting

Short-term load forecasting provides important foundation for the safety and economical operation of power system. With the fast development of modern electric power systems, the operation of power market requires high precision of short-term load forecasting for the minimal cost of power system operation. Currently there have been more studies in theory and complemented methods of load forecasting and obtained great achievement. New theory and new technology based on load forecasting researches have been developed continuously. As a new technology of data mining, support vector machine has been successfully applied in pattern recognition and regression problem, et al. Toward various factors of non-linear characteristics affecting power, there is a research about the method of short-term load forecasting based on support vector machine by using its advantages of non-linear processing and generating ability.The application profiles of the support vector machine in the field of short-term load forecasting are comprehensively summarized in this thesis. Starting from the principle of support vector machine and compared with artificial neural network method, the superiorities of the support vector machine method in the application of short-term load forecasting are elaborated. At the same time, some problems about the application of support vector machine, including data pre-processing, the constructing and selection of kernel function, and parameter optimization method, are analyzed in the thesis and the current solutions are provided respectively. In particular, for a series of support vector machine-based improvements and some mixed forecasting methods consisting of support vector machine with other algorithms, a comprehensive summary is given, from the perspective of the mechanism about support vector machine algorithm being applied into load forecasting, and the elevation of prediction accuracy and speed. Meantime, some key issues needing further discussion are put forward. Finally, this thesis summarizes the key issues about short-term load forecasting based on support vector machine, and gives some recommendations.In view of disadvantages of the single prediction, the exploration for comprehensive prediction has become a consensus among scholars. Therefore this paper adopts an effective ISODATA clustering analysis and process technology for the load data and combines ISODATA clustering algorithm with support vector machine. A new support vector machine method based on ISODATA clustering algorithm for short-term load forecasting is first presented in this thesis. Compared with the conventional support vector machine method, this method chooses training samples by ISODATA 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 application of the proposed method show the usefulness of this method, since both the precision and speed of load forecasting can be improved. The method based on ISODATA clustering algorithm for short-term load forecasting confirms the advantage of Cluster analysis in the load forecasting, also verifies the feasibility of using ISODATA algorithm to classify the load forecast data, and fully embodies the principle of the similarity of load forecasting.