基于PLS-LSSVM方法的湿法炼锌过程预测建模  

Forecasting modeling of zinc hydrometallurgy process based on PLS-LSSVM approach

作　　者：熊富强[1] 桂卫华[1] 阳春华[1] 王莎[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《仪器仪表学报》2011年第4期941-948,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金重点项目(60634020);国家杰出青年科学基金(61025015)资助项目

摘　　要：在湿法炼锌过程中,沉铁工序具有流程长、化学反应耦合强和物理沉降过程复杂的特点。针对这一特点,提出基于PLS-LSSVM的预测建模方法。PLS能有效处理复杂冶金生产过程中的非线性、多输入和数据相关性等特性。同时,针对建模过程中LS-SVM两个重要参数的优化选择,提出免疫文化差分进化算法。仿真结果表明,本文提出的基于PLS-LSSVM的预测模型能取得较好的预测结果,为解决沉铁过程铁渣品位实时检测提供了一种有效可行的方法。In Zinc hydrometallurgy process,the procedure of iron precipitation is a long flow,strong chemical reaction coupling and complex physical deposition process.Aiming at this characteristic,a forecasting modeling method based on PLS-LSSVM approach is put forward.The PLS can effectively handle the nonlinearity,multiple inputs and data correlation in complex metallurgical production process.Meanwhile,in order to optimize two important parameters of LS-SVM in the modeling process,immune cultural differential evolution algorithm（ICDEA） is proposed.Simulation results show that the proposed PLS-LSSVM method has achieved good model prediction result,and also can provide an effective and feasible method for iron residue grade real-time detection in the process of iron precipitation.

关 键 词：偏最小二乘法 最小二乘支持向量机 湿法炼锌 沉铁 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]