Study of the SMO algorithm based on data mining in shot-term power load forecasting model
Dongxiao Niu, Yongli Wang · 2008
A new methodology on the algorithm of sequential minimal optimization for the electric power system load was presented. In order to solve the problem that support vector machines can not deal with large scale data. First, this paper utilizes the advantage of data mining technology in processing large data and eliminating redundant information. The system mines the historical daily loading which has the same meteorological category as the forecasting day in order to compose data sequence with highly similar meteorological features, with this method it can decrease SVR training data; Second, this paper introduces the modified algorithm of sequential minimal optimization (SMO) to increase operational speed by use of a single threshold value. With this method it can decrease SVR training data and overcome the disadvantage of very large data and accelerates processing speed when constructing SVM model. The forecasted results are compared with those SVR employing QP optimization algorithm and BP artificial neural method, and it is shown that the presented forecasting method is more accurate and efficient.