基于PLS-LSSVM方法的湿法炼锌过程预测建模.pdfVIP

学兔兔 第32卷 第4期 仪 器 仪 表 学 报 V0l_32 No．4 2011年4月 Chinese Journal of Scientific Instrument Apr．2011 基于PLS．LSSVM方法的湿法炼锌过程预测建模术 熊富强，桂卫华，阳春华，王 莎 (中南大学信息科学与工程学院 长沙 410083) 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。 关键词：偏最小二乘法；最小二乘支持向量机；湿法炼锌；沉铁 中图分类号：TP273 文献标识码：A 国家标准学科分类代码：510．80 Forecasting modeling of zinc hydrometallurgy process based on PLS-LSSVM approach Xiong Fuqiang，Gui Weihua，Yang Chunhua，Wang Sha (School of Information ScienceEngineering，Central South University，Changsha 410083，China) Abstract：In Zinc hydrometallurgy process，the procedure of iron precipitation is a long flow，strong chemical reac— tion 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 parame— ters 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 precipita·- tion． Key words：partial least—squares；least squares support vector machine；zinc hydrometallurgy；iron precipitation 法和黄钾铁钒法3种工艺。本文所提及的是针铁矿法沉 1 引 言