摘要
混合核函数兼具了局部核函数和全局核函数的优点,并可通过权重因子调节它们对混合核函数的作用,取得较好的综合辨识效果。针对稀土萃取过程组分含量实时在线检测的难题,将基于混合核函数的支持向量机(SVM)算法用于稀土萃取过程组分含量建模。应用结果表明:基于混合核函数的组分含量软测量模型具有较好的模型拟合精度和满意的预测效果,能满足稀土萃取过程组分含量在线预估的要求。
Mixed kernel function, which has the characteristics of both local and global kernel function, can adjust its role through the weight factor to achieve a good synthesized identification effect. Aiming at the problem of real-time on- line measurement of the component content in rare earth counter-current extracting and separating process, the algorithm of SVM based on mixed kernel function is applied. The results of application indicate that the method based on mixed kernel function has both better fitting accuracy and satisfactory prediction effect, which can meet the online pre-estimating demands of component content in the rare-earth extracting process.
出处
《华东交通大学学报》
2010年第2期63-67,共5页
Journal of East China Jiaotong University
基金
国家自然科学基金项目(60864004)
江西省教育厅科学技术研究项目(GJJ10137)
华东交通大学科学技术研究项目(09DQ07)
关键词
稀土萃取
混合核函数
支持向量机
建模
rare-earth extraction
mixed kernel function
SVM
modeling
