基于神经网络补偿灰色预测误差的钴离子浓度预测研究  

Research of prediction in cobalt ions concentration based on a neural network compensating the error of grey forecast

作　　者：晏密英[1] 桂卫华[1] 王凌云[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《计算机与应用化学》2008年第7期805-808,共4页Computers and Applied Chemistry

基　　金：国家自然科学重点基金(60634020);博士点基金(20050533016).

摘　　要：某厂锌湿法冶炼三段净化过程中的Ⅱ段主要是通过过量添加锑盐和锌粉以除去有害杂质钴离子。本文从现场检测3个月的过程生产数据中,采用SPSS统计学软件深入分析了Ⅱ段净化工矿及其影响因素的相关性,得出了影响Ⅱ段后液钴离子浓度的主要因素,提出采用等维新息灰色预测方法预测Ⅱ段后液钴离子浓度,并采用神经网络补偿灰色预测的误差值。仿真和生产实践证明,该预测模型能够较好地预测Ⅱ段后液钴离子浓度值,从而为优化锑盐和锌粉添加量的操作起指导性作用。Three-grade purification process in zinc-hydrometallurgy is adopted in a plant. In the part-Ⅱ purification, it is mainly through the production of excessive addition of antimony salts and zinc to remove harmful impurities, cobalt ions. In this paper, production data were got from the scene of the process to deeply analysis the purification technics and also the relevant factors of the process Ⅱ （P-Ⅱ） with the usage of SPSS statistical software and then got the main factors of the part-Ⅱ. Under such a premise, a control method was brought forword in the article, with the adaption of the same-order newed grey prediction method to predict cobalt ion concentration of Part Ⅱ, and then, with a neural network, compensated the error of grey forecast. Simulation and production Practice has proved that the model can better predict the cobalt ion concentration values of P-Ⅱ of, so as to optimize the antimony salt and zinc addition and take a guiding role for the process operation.

关 键 词：净化 钴离子浓度 灰色预测 神经网络 

分 类 号：O6[理学—化学] TP393[自动化与计算机技术—计算机应用技术]