氧化铝蒸发工序出口浓度预测模型的建立  

The prediction Model of The Outlet Concentration of Alumina Evaporation System

作　　者：毕庆华[1] 桂卫华[1] 唐朝晖[1] 

机构地区：[1]中南大学信息科学与工程学院,410083

出　　处：《微计算机信息》2008年第36期194-196,共3页Control & Automation

基　　金：面向节能降耗的有色冶金过程控制若干理论与方法的研究(60634020)颁发部门:国家自然科学基金委基金申请人:桂卫华

摘　　要：氧化铝蒸发系统种分母液出口浓度是重要的出口质量指标。目前,在实际生产中种分蒸发工序多采用离线浓度检测,致使浓度控制严重滞后,影响了完成液的质量。由于BP神经网络具有良好的非线性拟合能力,因此,可以利用神经网络建立出口浓度回归模型。同时,为了使该模型具有较好的动态特性,对神经网络输出进行残差补偿,并用MATLAB对其进行了仿真研究和误差分析,实践表明该测量技术是可行的,其测量结果完全能满足工业生产的需要。The outlet concentration of Alumina evaporators system is an important outlet quality index. At present,the off-line method is universally adopted in real manufacture. This leads to severe lag in the concentration control and impacts in the quality of solution. As a result,the BP neutral network is used for building regression model of outlet concentration because of its favorable ca-pacity of unlinear data fitting. At the same time,the model of error compensation of linear regression is set up to modify the result of neutral network model in order to have favorable dynamic properties. Matlab is used for studying the model in simulated study and error analysis. In practice,it shows that this measuring technique is effective and its measuring result can fully meet the needs of in-dustrial production.

关 键 词：氧化铝蒸发系统 浓度预测模型 神经网络 残差补偿 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]