基于粗糙集理论和支持向量机的短期负荷预测研究-电力系统及其自动化专业论文.docxVIP

II II Abstract Economic and reliable operation of power systems depends significantly on the accuracy of the short-term load forecasting .The short-term load forecasting is very important for the unit optimal combination、economic dispatch and optimal power flow in the dispatching department, and it is especially significant to the power market. The more accurate the load forecasting is, the more efficient the usage of the units and the economic dispatch will be. With the development of the modern power system, the load forecasting need to be more accurate and rapid, and these requires make the load forecasting methods become more intelligentized and combination. This paper will research on one of the intelligentized methods—Support Vector Machine (SVM), find out its weakness and improve it, then built the combination model based on the SVM and rough set theory. The paper first researches the basis theory of SVM. The SVM model can more accurately forecast the future data by the structural risk minimization property. The sequential minimal optimization method can solve the quadratic programming problem of SVM. In order to analyse the inherent character of the load, the paper chooses the history load data of close time, and finds out the key attributes by the effective algorithm of attribute reduction in the rough set theory. On the other hand, the paper uses the load clustering analysis to improve the computing speed. We will choose the appropriate samples to make several SVM models, and the method can shorten the computing time. The combination model can exactly forecast the smooth load, then the paper continue to solve the load fast tracking problem after the mutable load and propose the similar day forecasting method. The method chooses the similar samples by the date type, weather type and the evening load level, then uses the combination model to forecast the load. At last the computation example shows that the method can solve the load fast tracking p