工程科学学报，第41卷，第8期：1052-1060,2019年8月 Chinese Journal of Engineering,Vol.41,No.8:1052-1060,August 2019 DOI:10.13374/j.issn2095-9389.2019.08.011;http://journals.ustb.edu.cn 基于PSO-RELM转炉冶炼终点锰含量预测模型 张壮)，曹玲玲)，林文辉)，孙建坤)，冯小明)，刘青) 1)北京科技大学钢铁冶金新技术国家重点实验室，北京1000832)新余钢铁集团有限公司，新余338001 区通信作者，E-mail:qiu@ustb.edu.cn 摘要分析了影响转炉治炼终点钢水中锰含量的因索，针对基于BP神经网络算法的转炉治炼终点锰含量预测模型存在的 收敛速度慢，预测精度低等问题，提出了一种基于极限学习机(ELM)算法建模的新思路，并引入正则化以及改进粒子群优化 算法(PSO),建立了基于改进粒子群算法优化的正则化极限学习机(PS0-RELM)的转炉终点锰含量预测模型：应用国内某 炼钢厂转炉实际生产数据对模型进行训练和验证，并与基于BP、ELM和RELM算法的三类模型进行比较.结果表明，采用P. S0-RELM方法构建的模型，锰含量预测误差在±0.025%范围内的命中率达到94%，均方误差为2.18×10~8，拟合优度R2为 0.72,上述三项指标均显著优于其他三类模型，此外，该模型还具有良好的泛化能力，对于转炉实际治炼过程具有一定的指导 意义. 关键词转炉：终点锰含量：改进粒子群算法：极限学习机：正则化极限学习机：预测模型 分类号TF723 Improved prediction model for BOF end-point manganese content based on IPSO-RELM method ZHANG Zhuang,CAO Ling-ling,LIN Wen-hui,SUN Jian-kun),FENG Xiao-ming?),LIU Qing 1)State Key Laboratory of Advanced Metallurgy,University of Science and Technology Beijing,Beijing 100083,China 2)Xinyu Iron Steel Group Co.,Ltd.,Xinyu 338001,China Corresponding author,E-mail:qliu@ustb.edu.cn ABSTRACT The basic oxygen furnace (BOF)steelmaking process,as the predominant steelmaking method used around the world, involves very complex physical and chemical phenomena such as multi-component reactions,multi-phase fluid dynamics,and high tem- perature.The main task of the BOF process is tailoring the temperature and melt components to meet the requirements of high-quality steel production.With the development of intelligent steelmaking,the prediction of the end-point manganese content is an extremely important task for the BOF process,and improving the level of control regarding the end-point of BOF steelmaking can reduce produc- tion costs and enhance efficiency.In this paper,the mechanism of the BOF steelmaking process and the factors influencing the end- point manganese content were analyzed.The control variables for predicting the end-point manganese content were also determined.To solve the problems of slow convergence,weak generalization ability,and low prediction accuracy in the prediction model established for the BP neural network,a new modeling concept based on an extreme learning machine (ELM)algorithm was proposed.By introducing regularization and improved particle swarm optimization(IPSO),a prediction model for the end-point manganese content in a converter based on improved particle swarm optimization and a regularized ELM (IPSO-RELM)was established.The paper then trained and verified the performance of these models with actual production data.A comparison of the performance of the proposed model with those of the prediction model of the BP neural network,the ELM model,and the RELM model reveals that the IPSO-RELM prediction model has the highest prediction accuracy and the best generalization performance.The hit ratio of the IPSO-RELM prediction model is 94% 收稿日期：2018-08-08 基金项目：江西省重点研发计划资助项目(20171ACE50020)工程科学学报,第 41 卷,第 8 期:1052鄄鄄1060,2019 年 8 月 Chinese Journal of Engineering, Vol. 41, No. 8: 1052鄄鄄1060, August 2019 DOI: 10. 13374 / j. issn2095鄄鄄9389. 2019. 08. 011; http: / / journals. ustb. edu. cn 基于 IPSO鄄鄄RELM 转炉冶炼终点锰含量预测模型 张 壮1) , 曹玲玲1) , 林文辉1) , 孙建坤1) , 冯小明2) , 刘 青1) 苣 1) 北京科技大学钢铁冶金新技术国家重点实验室, 北京 100083 2) 新余钢铁集团有限公司, 新余 338001 苣通信作者, E鄄mail: qliu@ ustb. edu. cn 摘 要 分析了影响转炉冶炼终点钢水中锰含量的因素,针对基于 BP 神经网络算法的转炉冶炼终点锰含量预测模型存在的 收敛速度慢,预测精度低等问题,提出了一种基于极限学习机(ELM)算法建模的新思路,并引入正则化以及改进粒子群优化 算法(IPSO),建立了基于改进粒子群算法优化的正则化极限学习机( IPSO鄄鄄RELM)的转炉终点锰含量预测模型;应用国内某 炼钢厂转炉实际生产数据对模型进行训练和验证,并与基于 BP、ELM 和 RELM 算法的三类模型进行比较. 结果表明,采用 IP鄄 SO鄄鄄RELM 方法构建的模型,锰含量预测误差在 依 0郾 025% 范围内的命中率达到 94% ,均方误差为 2郾 18 伊 10 - 8 ,拟合优度 R 2为 0郾 72,上述三项指标均显著优于其他三类模型,此外,该模型还具有良好的泛化能力,对于转炉实际冶炼过程具有一定的指导 意义. 关键词 转炉; 终点锰含量; 改进粒子群算法; 极限学习机; 正则化极限学习机;预测模型 分类号 TF723 收稿日期: 2018鄄鄄08鄄鄄08 基金项目: 江西省重点研发计划资助项目(20171ACE50020) Improved prediction model for BOF end鄄point manganese content based on IPSO鄄鄄RELM method ZHANG Zhuang 1) , CAO Ling鄄ling 1) , LIN Wen鄄hui 1) , SUN Jian鄄kun 1) , FENG Xiao鄄ming 2) , LIU Qing 1) 苣 1) State Key Laboratory of Advanced Metallurgy, University of Science and Technology Beijing, Beijing 100083, China 2) Xinyu Iron & Steel Group Co. , Ltd. , Xinyu 338001, China 苣Corresponding author, E鄄mail: qliu@ ustb. edu. cn ABSTRACT The basic oxygen furnace (BOF) steelmaking process, as the predominant steelmaking method used around the world, involves very complex physical and chemical phenomena such as multi鄄component reactions, multi鄄phase fluid dynamics, and high tem鄄 perature. The main task of the BOF process is tailoring the temperature and melt components to meet the requirements of high鄄quality steel production. With the development of intelligent steelmaking, the prediction of the end鄄point manganese content is an extremely important task for the BOF process, and improving the level of control regarding the end鄄point of BOF steelmaking can reduce produc鄄 tion costs and enhance efficiency. In this paper, the mechanism of the BOF steelmaking process and the factors influencing the end鄄 point manganese content were analyzed. The control variables for predicting the end鄄point manganese content were also determined. To solve the problems of slow convergence, weak generalization ability, and low prediction accuracy in the prediction model established for the BP neural network, a new modeling concept based on an extreme learning machine (ELM) algorithm was proposed. By introducing regularization and improved particle swarm optimization (IPSO), a prediction model for the end鄄point manganese content in a converter based on improved particle swarm optimization and a regularized ELM ( IPSO鄄鄄 RELM) was established. The paper then trained and verified the performance of these models with actual production data. A comparison of the performance of the proposed model with those of the prediction model of the BP neural network, the ELM model, and the RELM model reveals that the IPSO鄄鄄RELM prediction model has the highest prediction accuracy and the best generalization performance. The hit ratio of the IPSO鄄鄄RELM prediction model is 94%