基于动态主元分析和极限学习机的分解炉出口温度预测  

Prediction of Calciner Outlet Temperature Based on DPCA and ELM

作　　者：王祥民 董学平[1] 于广宇 WANG Xiang-min;DONG Xue-ping;YU Guang-yu(School of Electrical Engineering and Automation,Hefei University of Technology,Hefei 230009,China)

机构地区：[1]合肥工业大学电气与自动化工程学院

出　　处：《测控技术》2019年第12期35-39,76,共6页Measurement & Control Technology

基　　金：合肥工业大学产学研校企合作项目(w2015jskf0030)

摘　　要：在水泥生产过程中,为了应对分解炉结构的复杂性和影响出口温度变量的多样性,提出一种动态主元分析(Dynamic Principal Component Analysis,DPCA)与极限学习机(Extreme Learning Machine,ELM)相结合的数据驱动建模预测方法用来预测分解炉出口温度。通过采集的生产数据,提取影响出口温度变量的主元从而达到降维目的,将降维后的变量作为极限学习机的输入,分解炉出口温度作为极限学习机的输出。经极限学习机参数设置、训练、调整,得到出口温度预测模型。仿真验证结果表明,运用动态主元分析和极限学习机相结合的方法建立的分解炉出口温度预测模型具有良好的预测精度,且为后续出口温度的控制研究提供了依据,对水泥高效节能生产具有重要意义。In order to cope with the complexity of calciner structure and the variety of variables affecting the outlet temperature in the process of cement production,a data-driven modeling and prediction method combining dynamic principal component analysis(DPCA)and extreme learning machine(ELM)is proposed to predict the outlet temperature of calciner.The principal components affecting the outlet temperature variables were extracted through the production data collected to achieve the purpose of dimensionality reduction.The dimensionality reduced variables were taken as the input of the ELM,and the outlet temperature of calciner was taken as the output of the ELM.After setting,training and adjusting the ELM parameters,the outlet temperature prediction model was obtained.The simulation results show that the prediction model of outlet temperature established by the combination of DPCA and ELM has high prediction accuracy,and provides a basis for the follow-up research on outlet temperature control.It is of great significance for the cement production with high efficiency and energy saving.

关 键 词：分解炉出口温度 数据驱动建模 动态主元分析 降维 极限学习机 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]