基于SVM并考虑温度累积效应的电力短期负荷预测方法-计算机应用技术专业论文.docxVIP

Abstract Power short-term load forecasting is very important for energy optimal scheduling and planning of the electricity market. Now the electricity business from a planned economy monopoly model transform into a market economy competitive business model, and power energy supplying is tight, so an accurate load forecasting is crucial. The principle of similar-days algorithm is simple, and it is an effective way for improving the forecasting effect of short-term load forecasting by rationally choosing similar days. In recent years, Support Vector Machine (SVM) algorithm has been rapidly developed. It is a new machine learning algorithm, and it bases on the statistical learning theory. SVM is based on structural risk minimization, so it can achieve the minimum actual risk. It can be used to solve the problem of small sample, and has good generalization performance. It has been widely employed in the short term load forecasting. This paper is elaborated the forms of the accumulation effect of temperature based on fully analyzing the basic characteristics of short-term load. The summer accumulation effect of temperature is reflected by modifying the daily maximum temperature. The selected similar days，which are based on the modifying temperature, can be used to forecast the load in the next day. The work of the paper is following: The load characteristics of an area of Hubei province are detailed analysis from the time, meteorological and day types, etc. The advantages and disadvantages and modeling performance of three common forecasting methods (Time Series Analysis method, Neural Network method and Support Vector Machine method) are summarized, and the conclusion is that Support Vector Machine (SVM) method is more suitable for the study of short-term load forecasting. For the problem that the prediction accuracy isn’t ideal of Support Vector Machine method when the accumulation effect of temperature appears, a similar selection algorithm considering the accumulati