面向水泥分解炉工况预测的仿真代理模型构建研究  

Research on the construction of simulation agent model for the prediction of cement calciner working conditions

作　　者：丁一 何非 Ding Yi;He Fei(Nanjing Science and Technology University,Nanjing,210094,China)

机构地区：[1]南京理工大学机械工程学院工业工程系,江苏南京210094

出　　处：《水泥工程》2022年第6期1-5,9,共6页Cement Engineering

摘　　要：水泥分解炉在实际运行中存在部分关键工况参数难以通过测量取样直接获取的问题,目前主流的解决方法是直接数值模拟法,该方法存在响应时间长,不能并行计算多工况的缺点。本文提出构建基于Kriging模型的分解炉工况参数仿真代理模型,将分解炉分解率作为预测目标,对影响因子筛选并构建优化拉丁超立方抽样空间,通过有限元仿真计算获取响应值;利用Kriging插值模型对仿真数据进行拟合,得到Kriging仿真代理模型。模型检验结果表明,预测的R2达到了0.9178,代理模型预测结果与有限元仿真结果具有很好的一致性,可以实现对分解炉分解率的在线预测。In the actual operation of cement calciner,some key working condition parameters are difficult to obtain directly through measurement and sampling.At present,the mainstream solution is direct numerical simulation method,which has the disadvantage of long response time and can not calculate multiple working conditions in parallel.In this paper,a surrogate model for the simulation of calciner operating parameters based on Kriging model is proposed.The decomposition rate of the calciner is taken as the prediction target,and the influence factors are screened and optimized Latin hypercube sampling space is constructed.The response value is obtained through finite element simulation calculation;Kriging interpolation model is used to fit the simulation data to obtain Kriging simulation agent model.The model test results show that the predicted R2reaches 0.9178,and the proxy model prediction results are in good agreement with the finite element simulation results,which can realize the online prediction of the decomposition rate of the calciner.

关 键 词：水泥分解炉 数值模拟 KRIGING模型 

分 类 号：TQ172.6[化学工程—水泥工业]