摘要
支持向量机(SVM)已经成熟应用于非线性回归领域,并应用于工程费用量化分析。提出一种基于鲁棒支持向量回归机(RSVR)的量化新模型;通过优化支持向量机的损失函数,增加算法的鲁棒性,建立基于该原理的专用电力工程费用量化模型;并对样本数据添加人工噪声,提高样本数据的异常点,测试该模型的鲁棒性;算例分析表明:该文方法预测精度高、计算量小。
Support Vector Machine(SVM)has been applicated in the field of non-linear regression and in project cost quantification.A new electricial project quantitative model based on Robust Support Vector Regression(RSVR)was presented in this paper.By optimizing the loss function of the support vector machine,the robustness of the algorithm was increased.Based on this principle,a special electrical project cost evaluation model was established.And the anomaly of the sample data was increased to test its robustness by adding artificial noise.Example analysis results show that the proposed method is with high prediction accuracy,and less computation.
作者
王良缘
何轩
温步瀛
WANG Liang-yuan;HE Xuan;WEN Bu-ying(State Grid Fujian Electric Power Company Limited, Fuzhou 350003, China;School of Electrical Engineering and Automation,Fuzhou University, Minhou 350108, China)
出处
《电力科学与技术学报》
CAS
北大核心
2017年第1期78-83,共6页
Journal of Electric Power Science And Technology
基金
福建省自然科学基金(2013J01176)
关键词
专用工程费用
损失函数
支持向量机
费用量化模型
special project cost
loss function
support vector machine (SVM)
cost quantilization model