Application of Support Vector Machine in the Forecast of Tapping Aluminum Volume
Tapping aluminum volume is one of the important factors which influence technical and economic indexes in aluminum electrolysis. Based on the analysis of the theory of support vector regression, this paper proposes one kind of dual-feature weighted support vector machine, which treats training sample features with different weights by both the difference and the importance of the features, so it can reduce the features impact to the prediction result when the value of the feature is too large or the feature is unrelated to the result And combines with actual data, this paper establishes a prediction model of tapping aluminum volume. The experimental results indicate that the new algorithm obtains a better prediction.
support vector machine feature weighted forecast of tapping aluminum volume
Zhang Yanbin Li Jinhong Lin Manshan
College of Information Engineering North China University of Technology Beijing, China
国际会议
成都
英文
888-890
2010-12-17（万方平台首次上网日期，不代表论文的发表时间）