当前位置:高等教育资讯网  >  中国高校课件下载中心  >  大学文库  >  浏览文档

基于支持向量回归与极限学习机的高炉铁水温度预测

资源类别:文库,文档格式:PDF,文档页数:9,文件大小:1.27MB,团购合买
选取某4000 m3级别高炉2014年至2019年时间范围内的日平均数据,以铁水温度为目标函数,首先对铁水温度的特征参量进行线性与非线性相关性分析、特征选择与规范化处理,获取了显著影响铁水温度的正负相关性特征参量。在此基础上,基于支持向量回归与极限学习机两种算法对铁水温度构建预测模型,模型均可对铁水温度实现有效预测,基于支持向量回归算法构建的预测模型较优,预测平均绝对误差为4.33 ℃,±10 ℃误差范围内的命中率为94.0%。
点击下载完整版文档(PDF)

工程科学学报 Chinese Journal of Engineering 基于支持向量回归与极限学习机的高炉铁水温度预测 王振阳江德文王新东张建良刘征建赵宝军 Prediction of blast furnace hot metal temperature based on support vector regression and extreme learning machine WANG Zhen-yang.JIANG De-wen,WANG Xin-dong.ZHANG Jian-liang.LIU Zheng-jian,ZHAO Bao-jun 引用本文: 王振阳,江德文,王新东,张建良,刘征建,赵宝军.基于支持向量回归与极限学习机的高炉铁水温度预测工程科学学报, 2021,43(4:569-576.doi:10.13374.issn2095-9389.2020.05.28.001 WANG Zhen-yang,JIANG De-wen,WANG Xin-dong,ZHANG Jian-liang.LIU Zheng-jian,ZHAO Bao-jun.Prediction of blast furnace hot metal temperature based on support vector regression and extreme learning machine [Chinese Journal of Engineering, 2021,43(4:569-576.doi:10.13374f.issn2095-9389.2020.05.28.001 在线阅读View online::https:/doi.org/10.13374.issn2095-9389.2020.05.28.001 您可能感兴趣的其他文章 Articles you may be interested in 基于极限学习机ELMD的连铸坯质量预测 Quality prediction of the continuous casting bloom based on the extreme learning machine 工程科学学报.2018,40(7):815 https:1doi.org/10.13374j.issn2095-9389.2018.07.007 基于改进的支持向量回归机算法的磁记忆定量化缺陷反演 Metal magnetic memory quantitative inversion of defects based onoptimized support vector machine regression 工程科学学报.2018,40(9%:1123 https:1doi.org/10.13374.issn2095-9389.2018.09.014 函数型数据分析与优化极限学习机结合的弹药传输机械臂参数辨识 Parameter identification of a shell transfer arm using FDA and optimized ELM 工程科学学报.2017,394:611htps:/doi.org10.13374 /j.issn2095-9389.2017.04.017 基于机器学习的北京市PM2.5浓度预测模型及模拟分析 Machine-learning-based model and simulation analysis of PM2.5 concentration prediction in Beijing 工程科学学报.2019,41(3:401 https:/1doi.org10.13374j.issn2095-9389.2019.03.014 基于多目标支持向量机的ADHD分类 ADHD classification based on a multi-objective support vector machine 工程科学学报.2020,42(4:441htps:/oi.org10.13374.issn2095-9389.2019.09.12.007 多模态学习方法综述 A survey of multimodal machine learning 工程科学学报.2020,42(5:557 https:/doi.org10.13374j.issn2095-9389.2019.03.21.003

基于支持向量回归与极限学习机的高炉铁水温度预测 王振阳 江德文 王新东 张建良 刘征建 赵宝军 Prediction of blast furnace hot metal temperature based on support vector regression and extreme learning machine WANG Zhen-yang, JIANG De-wen, WANG Xin-dong, ZHANG Jian-liang, LIU Zheng-jian, ZHAO Bao-jun 引用本文: 王振阳, 江德文, 王新东, 张建良, 刘征建, 赵宝军. 基于支持向量回归与极限学习机的高炉铁水温度预测[J]. 工程科学学报, 2021, 43(4): 569-576. doi: 10.13374/j.issn2095-9389.2020.05.28.001 WANG Zhen-yang, JIANG De-wen, WANG Xin-dong, ZHANG Jian-liang, LIU Zheng-jian, ZHAO Bao-jun. Prediction of blast furnace hot metal temperature based on support vector regression and extreme learning machine [J]. Chinese Journal of Engineering, 2021, 43(4): 569-576. doi: 10.13374/j.issn2095-9389.2020.05.28.001 在线阅读 View online: https://doi.org/10.13374/j.issn2095-9389.2020.05.28.001 您可能感兴趣的其他文章 Articles you may be interested in 基于极限学习机(ELM)的连铸坯质量预测 Quality prediction of the continuous casting bloom based on the extreme learning machine 工程科学学报. 2018, 40(7): 815 https://doi.org/10.13374/j.issn2095-9389.2018.07.007 基于改进的支持向量回归机算法的磁记忆定量化缺陷反演 Metal magnetic memory quantitative inversion of defects based onoptimized support vector machine regression 工程科学学报. 2018, 40(9): 1123 https://doi.org/10.13374/j.issn2095-9389.2018.09.014 函数型数据分析与优化极限学习机结合的弹药传输机械臂参数辨识 Parameter identification of a shell transfer arm using FDA and optimized ELM 工程科学学报. 2017, 39(4): 611 https://doi.org/10.13374/j.issn2095-9389.2017.04.017 基于机器学习的北京市PM2.5浓度预测模型及模拟分析 Machine-learning-based model and simulation analysis of PM2.5 concentration prediction in Beijing 工程科学学报. 2019, 41(3): 401 https://doi.org/10.13374/j.issn2095-9389.2019.03.014 基于多目标支持向量机的ADHD分类 ADHD classification based on a multi-objective support vector machine 工程科学学报. 2020, 42(4): 441 https://doi.org/10.13374/j.issn2095-9389.2019.09.12.007 多模态学习方法综述 A survey of multimodal machine learning 工程科学学报. 2020, 42(5): 557 https://doi.org/10.13374/j.issn2095-9389.2019.03.21.003

工程科学学报.第43卷,第4期:569-576.2021年4月 Chinese Journal of Engineering,Vol.43,No.4:569-576,April 2021 https://doi.org/10.13374/j.issn2095-9389.2020.05.28.001;http://cje.ustb.edu.cn 基于支持向量回归与极限学习机的高炉铁水温度预测 王振阳)区,江德文),王新东),张建良1,3),刘征建),赵宝军) 1)北京科技大学治金与生态工程学院,北京1000832)河钢集团有限公司钢铁技术研究总院,石家庄0500233)昆士兰大学化学工程学 院,圣卢西亚QLD4072 ☒通信作者.E-mail:wangzhenyang(@ustb.edu.cn 摘要选取某4000m3级别高炉2014年至2019年时间范围内的日平均数据,以铁水温度为目标函数.首先对铁水温度的 特征参量进行线性与非线性相关性分析、特征选择与规范化处理,获取了显著影响铁水温度的正负相关性特征参量.在此基 础上,基于支持向量回归与极限学习机两种算法对铁水温度构建预测模型,模型均可对铁水温度实现有效预测,基于支持向 量回归算法构建的预测模型较优,预测平均绝对误差为4.33℃,±10℃误差范围内的命中率为94.0%. 关键词大数据:机器学习:支持向量回归:极限学习机:铁水温度 分类号TF543.1 Prediction of blast furnace hot metal temperature based on support vector regression and extreme learning machine WANG Zhen-yang JIANG De-wen.WANG Xin-dong.ZHANG Jian-liang LIU Zheng-jian,ZHAO Bao-jun 1)School of Metallurgical and Ecological Engineering,University of Science and Technology Beijing,Beijing 100083,China 2)Iron and Steel Technology Research Institute,Hegang Group Co.Ltd.,Shijiazhuang 050023,China 3)School of Chemical Engineering,The University of Queensland,St Lucia QLD 4072,Australia Corresponding author,E-mail:wangzhenyang@ustb.edu.cn ABSTRACT The hot metal temperature is a key process parameter for blast furnace(BF)ironmaking that reflects the quality of hot metal,the thermal state of BF hearth,the energy utilization efficiency of BF,and many other information.Prediction of the hot metal temperature in the next smelting cycle will be helpful in gaining a better understanding of the change trend of hot metal quality and BF smelting status in time.With this,corresponding operational measures can be conducted to maintain the BF stable and smooth state,high production,and low consumption.Nowadays,big data technology has made considerable progress toward a more accurate and faster collection,storage,transmission,query,analysis,and integration of mass data,providing a good data foundation for data-driven machine learning models.In addition,with the substantial increase in computer calculation speed and the significant development of algorithms, the prediction accuracy of deep machine learning models has noticeably improved.The development of these technologies provides feasibility for the prediction of important indicators under complex industrial conditions.Based on the data produced from a 4000-mBF in a large span time range(2014-2019)and daily time dimension,this paper considered hot metal temperature as the objective function. First,the characteristic parameters of hot metal temperature were processed by linear and nonlinear correlation analysis,feature selection,and normalization methods.Then,the positive and negative correlation characteristic parameters that have a significant influence on the temperature of the hot metal were obtained.Finally,prediction models of hot metal temperature were established based on two algorithms of support vector regression and extreme learning machine.Although both the algorithms can achieve effective 收稿日期:202005-28 基金项目:中国博士后科学基金面上资助项目(2019M650490)

基于支持向量回归与极限学习机的高炉铁水温度预测 王振阳1) 苣,江德文1),王新东2),张建良1,3),刘征建1),赵宝军3) 1) 北京科技大学冶金与生态工程学院,北京 100083    2) 河钢集团有限公司钢铁技术研究总院,石家庄 050023    3) 昆士兰大学化学工程学 院,圣卢西亚 QLD 4072 苣通信作者,E-mail: wangzhenyang@ustb.edu.cn 摘    要    选取某 4000 m 3 级别高炉 2014 年至 2019 年时间范围内的日平均数据,以铁水温度为目标函数,首先对铁水温度的 特征参量进行线性与非线性相关性分析、特征选择与规范化处理,获取了显著影响铁水温度的正负相关性特征参量. 在此基 础上,基于支持向量回归与极限学习机两种算法对铁水温度构建预测模型,模型均可对铁水温度实现有效预测,基于支持向 量回归算法构建的预测模型较优,预测平均绝对误差为 4.33 ℃,±10 ℃ 误差范围内的命中率为 94.0%. 关键词    大数据;机器学习;支持向量回归;极限学习机;铁水温度 分类号    TF543.1 Prediction  of  blast  furnace  hot  metal  temperature  based  on  support  vector  regression and extreme learning machine WANG Zhen-yang1) 苣 ,JIANG De-wen1) ,WANG Xin-dong2) ,ZHANG Jian-liang1,3) ,LIU Zheng-jian1) ,ZHAO Bao-jun3) 1) School of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China 2) Iron and Steel Technology Research Institute, Hegang Group Co. Ltd., Shijiazhuang 050023, China 3) School of Chemical Engineering, The University of Queensland, St Lucia QLD 4072, Australia 苣 Corresponding author, E-mail: wangzhenyang@ustb.edu.cn ABSTRACT    The hot metal temperature is a key process parameter for blast furnace (BF) ironmaking that reflects the quality of hot metal, the thermal state of BF hearth, the energy utilization efficiency of BF, and many other information. Prediction of the hot metal temperature in the next smelting cycle will be helpful in gaining a better understanding of the change trend of hot metal quality and BF smelting status in time. With this, corresponding operational measures can be conducted to maintain the BF stable and smooth state, high production, and low consumption. Nowadays, big data technology has made considerable progress toward a more accurate and faster collection, storage, transmission, query, analysis, and integration of mass data, providing a good data foundation for data-driven machine learning models. In addition, with the substantial increase in computer calculation speed and the significant development of algorithms, the  prediction  accuracy  of  deep  machine  learning  models  has  noticeably  improved.  The  development  of  these  technologies  provides feasibility for the prediction of important indicators under complex industrial conditions. Based on the data produced from a 4000-m3 BF in a large span time range (2014–2019) and daily time dimension, this paper considered hot metal temperature as the objective function. First,  the  characteristic  parameters  of  hot  metal  temperature  were  processed  by  linear  and  nonlinear  correlation  analysis,  feature selection,  and  normalization  methods.  Then,  the  positive  and  negative  correlation  characteristic  parameters  that  have  a  significant influence on the temperature of the hot metal were obtained. Finally, prediction models of hot metal temperature were established based on  two  algorithms  of  support  vector  regression  and  extreme  learning  machine.  Although  both  the  algorithms  can  achieve  effective 收稿日期: 2020−05−28 基金项目: 中国博士后科学基金面上资助项目(2019M650490) 工程科学学报,第 43 卷,第 4 期:569−576,2021 年 4 月 Chinese Journal of Engineering, Vol. 43, No. 4: 569−576, April 2021 https://doi.org/10.13374/j.issn2095-9389.2020.05.28.001; http://cje.ustb.edu.cn

570 工程科学学报,第43卷,第4期 prediction,results from support vector regression are better at an average absolute error of 4.33 C and a hit rate of 94.0%(+10C). KEY WORDS big data;machine learning;support vector regression;extreme learning machine;hot metal temperature 2000年以来,我国生铁产量快速增加,并于 基础的模型在炼铁领域应用较为广泛61网,已成 2019年全年产量达到8.06亿吨,约占当年世界生 功实现对铁水硅含量和高炉故障的预判0,此 铁产量的63%.能耗方面,我国钢铁行业能耗约占 外,极限学习机(Extreme learning machine,简称 工业总能耗的16%,而炼铁工序又占钢铁全流程 ELM)因隐藏神经元的参数无需调整,而获得快速 能耗的70%左右.在此背景下,高炉作为炼铁工序 的训练速度和良好的工业适应度2-2]因此,本研 的主反应器,其稳定顺行对我国钢铁工业的节能 究以支持向量回归和极限学习机算法为基础,在 降耗与高效生产尤为重要.然而高炉原燃料波动 日平均数据的时间维度下,对铁水温度进行模型 造成的炉型改变,以及设备破损导致的不均匀传 学习和预测研究,期望在拓宽时间维度的基础上 热等多种内外因素使高炉炉况波动时有发生,给 获得工艺可接受的铁水温度预测准确度,为高炉 炼铁工序的增产降耗带来影响- 在线实际应用提供基础模型支撑 铁水温度是高炉炼铁的一项重要参数,不仅 可以反映当前铁水本身质量与高炉炉缸热状态, 1 理论与方法 且通过其变化趋势和幅度也能预判高炉运转方 1.1SVR算法 向.因此,若能通过当前时刻的各项治炼参数预测 回归预测模型的构建一般是基于训练样本集 一定时间后的铁水温度变化趋势与幅度,则可以 D={cm,ym)}而获得形如x)=wx+b模型中w和 提前采取应对凋控措施,减少炉况波动,稳定高炉 b等未知参量,传统回归模型通常是直接以预测 热制度,从而促进高炉稳定顺行与炼铁工序节能 值x)与测量值y之间的差别计算损失,并通过优 降耗B3- 化损失函数以降低误差,使x)与y之间的差别 基于此,国内外研究学者对铁水温度展开了 (损失)减小,损失函数(0如式(1)所示 多种构建机制条件下的机器学习与模型预测研 究,取得了一定进展.崔桂梅等分别采用模糊神经 J0=∑hex)-yw2 (1) 网络与分布式神经网络网构建铁水温度预测模 =1 型,使铁水温度预测误差在±10℃内的准确度达 其中,m为样本量,x为第i个训练样本特征值组 到90%.石琳等9基于样条变换的非线性偏最小 合,为为第i个训练样本测量值,hc)为第i个训 二乘回归方法,控制铁水温度预测值与测量值的 练样本特征值组合预测函数,0为选定的某个预测 相对误差小于0.1l.Zhang等lo以均方根误差和 函数标号 相关系数为标准,对比分析了3种深度学习算法 对于SVR算法,可允许x)与y之间存在e的 与7种浅层学习算法对高炉铁水温度的预测效 偏差,即仅当x)pε才计算损失.若训练样本是 果,认为浅层神经网络(SNN)对当前铁温预测精 线性可分的,则在样本空间中,可通过式(2)描述 度较高,且计算资源消耗较低,适合在线反馈.此 其划分出的超平面 外,Zhang等山还通过整合单一模式树构建集合 wx+b=0 (2) 模式树,并据此进一步将铁水温度预测均方根误 其中,w为法向量,b为位移项.若将x从低维空间 差降低至7.33.基于多元自适应回归样条(MARS) 算法和滚动时域估计(MHE),Diaz等和Hashi- 映射到高维空间的特征向量表示为(x),则高维 moto等]对铁水温度预测的平均绝对误差分别 空间划分的超平面模型如式(3)所示 为11.2℃和11.6℃.通过结合粒子群算法(APS0), f(x)=wD(x)+b (3) Su等和Zhao等分别以极限学习机(ELM)和 此时,样本空间任意点x到超平面(e,b)的距 最小二乘支持向量机(LS-SVM)作为基础算法预 离r与间隔宽度y分别如式(4)、(5)所示 测铁水温度,模型取得较好预测精度和泛化性能 r=mT)+以 (4) 基于数据驱动的预测模型还包括基于时间序 hell 列的自回归算法、贝叶斯网络算法等,其中以支持 28 =Iwill (5) 向量回归(Support vector regression,.简称SVR)为

prediction, results from support vector regression are better at an average absolute error of 4.33 °C and a hit rate of 94.0% (±10 °C). KEY WORDS    big data;machine learning;support vector regression;extreme learning machine;hot metal temperature 2000 年以来,我国生铁产量快速增加,并于 2019 年全年产量达到 8.06 亿吨,约占当年世界生 铁产量的 63%. 能耗方面,我国钢铁行业能耗约占 工业总能耗的 16%,而炼铁工序又占钢铁全流程 能耗的 70% 左右. 在此背景下,高炉作为炼铁工序 的主反应器,其稳定顺行对我国钢铁工业的节能 降耗与高效生产尤为重要. 然而高炉原燃料波动 造成的炉型改变,以及设备破损导致的不均匀传 热等多种内外因素使高炉炉况波动时有发生,给 炼铁工序的增产降耗带来影响[1−2] . 铁水温度是高炉炼铁的一项重要参数,不仅 可以反映当前铁水本身质量与高炉炉缸热状态, 且通过其变化趋势和幅度也能预判高炉运转方 向. 因此,若能通过当前时刻的各项冶炼参数预测 一定时间后的铁水温度变化趋势与幅度,则可以 提前采取应对调控措施,减少炉况波动,稳定高炉 热制度,从而促进高炉稳定顺行与炼铁工序节能 降耗[3−6] . 基于此,国内外研究学者对铁水温度展开了 多种构建机制条件下的机器学习与模型预测研 究,取得了一定进展. 崔桂梅等分别采用模糊神经 网络[7] 与分布式神经网络[8] 构建铁水温度预测模 型,使铁水温度预测误差在±10 ℃ 内的准确度达 到 90%. 石琳等[9] 基于样条变换的非线性偏最小 二乘回归方法,控制铁水温度预测值与测量值的 相对误差小于 0.11. Zhang 等[10] 以均方根误差和 相关系数为标准,对比分析了 3 种深度学习算法 与 7 种浅层学习算法对高炉铁水温度的预测效 果,认为浅层神经网络(SNN)对当前铁温预测精 度较高,且计算资源消耗较低,适合在线反馈. 此 外 ,Zhang 等[11] 还通过整合单一模式树构建集合 模式树,并据此进一步将铁水温度预测均方根误 差降低至 7.33. 基于多元自适应回归样条(MARS) 算法和滚动时域估计(MHE) ,Diaz 等[12] 和 Hashi￾moto 等[13] 对铁水温度预测的平均绝对误差分别 为 11.2 ℃ 和 11.6 ℃. 通过结合粒子群算法(APSO), Su 等[14] 和 Zhao 等[15] 分别以极限学习机(ELM)和 最小二乘支持向量机(LS-SVM)作为基础算法预 测铁水温度,模型取得较好预测精度和泛化性能. 基于数据驱动的预测模型还包括基于时间序 列的自回归算法、贝叶斯网络算法等,其中以支持 向量回归(Support vector regression,简称 SVR)为 基础的模型在炼铁领域应用较为广泛[16−18] ,已成 功实现对铁水硅含量[19] 和高炉故障的预判[20] . 此 外 ,极限学习机 ( Extreme  learning  machine, 简 称 ELM)因隐藏神经元的参数无需调整,而获得快速 的训练速度和良好的工业适应度[21−23] . 因此,本研 究以支持向量回归和极限学习机算法为基础,在 日平均数据的时间维度下,对铁水温度进行模型 学习和预测研究,期望在拓宽时间维度的基础上 获得工艺可接受的铁水温度预测准确度,为高炉 在线实际应用提供基础模型支撑. 1    理论与方法 1.1    SVR 算法 回归预测模型的构建一般是基于训练样本集 D={(xm, ym)}而获得形 如 f(x)=w T x+b 模 型 中 w 和 b 等未知参量,传统回归模型通常是直接以预测 值 f(x) 与测量值 y 之间的差别计算损失,并通过优 化损失函数以降低误差,使 f(x) 与 y 之间的差别 (损失)减小,损失函数 J(θ) 如式(1)所示. J(θ)= 1 2 ∑m i=1 (hθ(xi)−yi) 2 (1) 其中,m 为样本量,xi 为第 i 个训练样本特征值组 合,yi 为第 i 个训练样本测量值,hθ (xi ) 为第 i 个训 练样本特征值组合预测函数,θ 为选定的某个预测 函数标号. 对于 SVR 算法,可允许 f(x) 与 y 之间存在ε的 偏差,即仅当|f(x)−y|>ε 才计算损失. 若训练样本是 线性可分的,则在样本空间中,可通过式(2)描述 其划分出的超平面. w T x+b = 0 (2) 其中,w 为法向量,b 为位移项. 若将 x 从低维空间 映射到高维空间的特征向量表示为 Φ(x),则高维 空间划分的超平面模型如式(3)所示. f(x) = w TΦ(x)+b (3) 此时,样本空间任意点 x 到超平面(w,b)的距 离 r 与间隔宽度 γ 分别如式(4)、(5)所示. r = |w TΦ(x)+b| ||w|| (4) γ = 2ε ||w|| (5) · 570 · 工程科学学报,第 43 卷,第 4 期

王振阳等:基于支持向量回归与极限学习机的高炉铁水温度预测 571. 以x)为中心,构建宽度为2ε的间隔带,若训 影响.本文依据冶金原理以及现场监测条件,选取 练样本落入此间隔带,则认为是预测正确.进一步 国内某4000m3级别高炉2014年至2019年时间范 地,将核函数引入SVR后如式(6)所示 围内22项特征参量的日平均数据构成初选数据 集,特征参量与实际数据样本分别如表1和图1所示 f(x)= (ai-ai).k(xi,x)+b (6) i=l 表1铁水温度预测的初选特征参量 式中: Table 1 Primary data items for hot metal temperature prediction b=yi+8- >(ai-ai)xIx Operating parameters State parameters (7) Blast volume Volume utilization coefficient 其中,kx,x)为核函数,拉格朗日乘子a:≥0、a,≥ Blast pressure Synthetic load 0,其中核函数的选择至关重要,通过优化选择适 Blast temperature Gas utilization efficiency 用的核函数(例如:径向基核函数、线性核函数、 Blast velocity energy Daily hot metal production 多项式核函数或Sigmoid核函数),从而最终确定 Coke rate Pressure difference 基于SVR的预测模型 Coal injection rate Permeability index 1.2ELM算法 Nut coke rate Bosh gas volume ELM是用于单层前馈神经网络的训练学习方 Fuel rate Bosh gas index 法,单层前馈网络拥有三层神经元,即输入层、隐 Oxygen enrichment Cooling water temperature difference 藏层和输出层.对于含有N个样本的数据集(: Pulverized coal injection per hour Current hot metal temperature ),其单层前馈神经网络数学模型如式(8)所示 Theoretical combustion temperature Hot metal Si content Bigi(xk)= Big(wi,bi,xk)=tk,k =1,2,....N 图1所示铁水温度的初选特征参量共计22 (8) 项.特征参量过多将增加模型计算复杂度,并大幅 其中,xx,x2,…,x】表示第k个样本的N维 提高计算时间,因此需要对主要特征参量进行提 特征,=,2,…,]表示对应的目标向量,B,是 取.本文依据冶金原理、Pearson相关系数以及 输出权重矩阵,用于连接第i个隐藏和输出节点, Spearman相关系数对初选特征参量进一步筛选, gw,bnx)为非线性分段连续函数,极限能力满 以减少后续模型训练与预测过程中的无效计算, 足ELM定理,w,和b,为随机确定的模型参数.因 降低模型的计算复杂度,以适应工业生产条件 而,式(8)可写成隐含输出矩阵形式,如式(9)、 Pearson与Spearman相关系数计算方法分别如式 (10)所示,其中,H为隐含层输出矩阵,B为输出权 (12)与(13)所示,式中,X为特征向量,Y为目标向 重矩阵,T为目标矩阵 量,Pxn和R分别为X与Y的Pearson和Spearman HB=T (9) 相关系数,cov(X)为X和Y的协方差,Var()与 Var()为X与Y的样本方差,N为样本数,RX)与 g(w1·x1+b1) … g(wN·r1+bR) H(W.B.X)= R(Y)分别为元素X和Y,在X和Y中的排序.经计 g(w1·xN+bI) …g(wNxN+b)JNxN 算,初选特征参量与铁水温度的相关系数计算结 (10) 果如图2所示 根据最小二乘法理论,结合奇异值分解方法, cov(X,Y) P(X.)= (12) ELM的解可表示为式(11) Var(X)Var(Y) B=(HH)HTT (11) N 6×[R(X)-R(Y2 2铁水温度预测模型 R=1- (13) N×(W2-1) 2.1特征参量提取与数据标准化 由图2可知,对铁水温度发展走向的影响因素 高炉生产过程中会产生两类参数,即人为调 较多,既包含影响高炉直接与间接还原度的炉腹 控的主动操作参数以及反馈炉况的被动状态参 煤气指数、煤气利用率等,也包含影响高炉物料与 数,均会对铁水温度的走向变化产生不同程度的 热量平衡的炉料结构、水温差等.具体来看,对后

以 f(x) 为中心,构建宽度为 2ε 的间隔带,若训 练样本落入此间隔带,则认为是预测正确. 进一步 地,将核函数[24] 引入 SVR 后如式(6)所示. f(x) = ∑m i=1 (ˆai −ai)· k(xi , x)+b (6) 式中: b = yi +ε− ∑m i=1 (ˆai −ai)x T i x (7) 其中,k(xi,x) 为核函数,拉格朗日乘子aˆi ⩾ 0、ai ≥ 0,其中核函数的选择至关重要,通过优化选择适 用的核函数(例如:径向基核函数、线性核函数、 多项式核函数或 Sigmoid 核函数),从而最终确定 基于 SVR 的预测模型. 1.2    ELM 算法 ELM 是用于单层前馈神经网络的训练学习方 法,单层前馈网络拥有三层神经元,即输入层、隐 藏层和输出层. 对于含有 N 个样本的数据集 (xk, tk ),其单层前馈神经网络数学模型如式(8)所示. ∑ N˜ i=1 βigi(xk) = ∑ N˜ i=1 βig(wi ,bi , xk) = tk, k = 1,2,··· ,N (8) , ··· , , ··· , 其中, xk=[xk1, xk2 xkn] 表示第 k 个样本的 N 维 特征,tk=[tk1,tk2 tkn] 表示对应的目标向量,βi 是 输出权重矩阵,用于连接第 i 个隐藏和输出节点, g(wi,bi,xk ) 为非线性分段连续函数,极限能力满 足 ELM 定理,wi 和 bi 为随机确定的模型参数. 因 而,式( 8)可写成隐含输出矩阵形式,如式( 9) 、 (10)所示,其中,H 为隐含层输出矩阵,β 为输出权 重矩阵,T 为目标矩阵. Hβ = T (9) H(W,B,X) =   g(w1 · x1 +b1) ... g(wN˜ · x1 +bN˜ ) . . . . . . . . . g(w1 · xN +b1) ... g(wN˜ · xN +bN˜ )   N×N˜ (10) 根据最小二乘法理论,结合奇异值分解方法, ELM 的解可表示为式(11). βˆ = (H TH) −1H TT (11) 2    铁水温度预测模型 2.1    特征参量提取与数据标准化 高炉生产过程中会产生两类参数,即人为调 控的主动操作参数以及反馈炉况的被动状态参 数,均会对铁水温度的走向变化产生不同程度的 影响. 本文依据冶金原理以及现场监测条件,选取 国内某 4000 m 3 级别高炉 2014 年至 2019 年时间范 围内 22 项特征参量的日平均数据构成初选数据 集,特征参量与实际数据样本分别如表 1 和图 1 所示. 表 1 铁水温度预测的初选特征参量 Table 1   Primary data items for hot metal temperature prediction Operating parameters State parameters Blast volume Volume utilization coefficient Blast pressure Synthetic load Blast temperature Gas utilization efficiency Blast velocity energy Daily hot metal production Coke rate Pressure difference Coal injection rate Permeability index Nut coke rate Bosh gas volume Fuel rate Bosh gas index Oxygen enrichment Cooling water temperature difference Pulverized coal injection per hour Current hot metal temperature Theoretical combustion temperature Hot metal Si content 图 1 所示铁水温度的初选特征参量共计 22 项. 特征参量过多将增加模型计算复杂度,并大幅 提高计算时间,因此需要对主要特征参量进行提 取. 本文依据冶金原理、 Pearson 相关系数以及 Spearman 相关系数对初选特征参量进一步筛选, 以减少后续模型训练与预测过程中的无效计算, 降低模型的计算复杂度,以适应工业生产条件. Pearson 与 Spearman 相关系数计算方法分别如式 (12)与(13)所示,式中,X 为特征向量,Y 为目标向 量,ρ(X, Y) 和 Rs 分别为 X 与 Y 的 Pearson 和 Spearman 相关系数,cov(X,Y) 为 X 和 Y 的协方差,Var(X) 与 Var(Y) 为 X 与 Y 的样本方差,N 为样本数,R(Xi ) 与 R(Yi ) 分别为元素 Xi 和 Yi 在 X 和 Y 中的排序. 经计 算,初选特征参量与铁水温度的相关系数计算结 果如图 2 所示. ρ(X,Y)= cov(X,Y) √ Var(X) √ Var(Y) (12) Rs = 1− 6× ∑ N i=1 [R(Xi)−R(Yi)]2 N ×(N2 −1) (13) 由图 2 可知,对铁水温度发展走向的影响因素 较多,既包含影响高炉直接与间接还原度的炉腹 煤气指数、煤气利用率等,也包含影响高炉物料与 热量平衡的炉料结构、水温差等. 具体来看,对后 王振阳等: 基于支持向量回归与极限学习机的高炉铁水温度预测 · 571 ·

572 工程科学学报,第43卷,第4期 9000 1) (12) 6000 aunjoA uonezi jn (2) 4.0 (13) 0.3 2.5 1300 (3) (14) 1100 4) 12000 0000 (15) 6000 5000 (5 400 0.25 (16) 0.20 300 0.15 150 (6 20 100 50 10 80 2×10 40 uin]o 1×10 600 (19) 500 60 50 (9) M u00 (20) 2 60 (10) 1520 (21) 30 0 1440 2400 (11) 1.0 (22) 0.5 2100 0 0 200400 6008001000120014001600 0 2004006008001000120014001600 Sample number Sample number 图1铁水温度初选特征参量样本散点图 Fig.I Scatter plot of the primary characteristic parameters of hot metal temperature 续铁水温度存在显著正相关关系的特征参量主要 的影响.零-均值规范可以减小噪声点的影响,且 为当前铁水温度、煤气利用率、铁水硅含量、利用 不受特征参量最大值和最小值的限制,在当前数 系数、铁水日产量和综合负荷等,存在显著负相关 据标准化中使用较为广泛,其计算公式如式(14) 关系的特征参量为炉腹煤气指数、水温差、焦丁 所示,其中:为原始数据的均值,σ为原始数据的 比、燃料比、焦比和透气性等.通过预实验计算 标准差,‘为标准化处理后的样本值 并综合考量计算复杂度与预测精准度,最终选取 2=-) (14) 如表2所示特征参量进行模型拟合,其中1~4项 为操作参数,5~14项为状态参数,删除相关度较 2.2铁水温度模型预测与结果表征 低的冗余特征参量 铁水温度特征参量提取及数据标准化处理 本文使用零-均值规范法对数据进行标准化 后,形成的数据集将用于模型的训练与测试.数据 处理,以消除特征参量之间量纲和取值范围差异 集中随机抽取80%的数据用于支持向量回归与极

续铁水温度存在显著正相关关系的特征参量主要 为当前铁水温度、煤气利用率、铁水硅含量、利用 系数、铁水日产量和综合负荷等,存在显著负相关 关系的特征参量为炉腹煤气指数、水温差、焦丁 比、燃料比、焦比和透气性等. 通过预实验计算, 并综合考量计算复杂度与预测精准度,最终选取 如表 2 所示特征参量进行模型拟合,其中 1~4 项 为操作参数,5~14 项为状态参数,删除相关度较 低的冗余特征参量. 本文使用零–均值规范法对数据进行标准化 处理,以消除特征参量之间量纲和取值范围差异 z¯ 的影响. 零–均值规范可以减小噪声点的影响,且 不受特征参量最大值和最小值的限制,在当前数 据标准化中使用较为广泛,其计算公式如式(14) 所示,其中: 为原始数据的均值,σ 为原始数据的 标准差,z *为标准化处理后的样本值. z ∗ = (z−z¯) σ (14) 2.2    铁水温度模型预测与结果表征 铁水温度特征参量提取及数据标准化处理 后,形成的数据集将用于模型的训练与测试. 数据 集中随机抽取 80% 的数据用于支持向量回归与极 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) Blast volume/ (m3·min−1 ) Bosh gas volume/ (m3·min−1 ) Bosh gas index Cooling water temperature difference/ ℃ Current hot metal temperature/ ℃ Hot metal Si content/ % Blast pressure/ MPa Blast temperature/ ℃ Blast velocity energy/J Coke rate/ (kg·t−1 ) Coal injection rate/(kg·t−1 ) Nut coke rate/(kg·t−1 ) Pulverized coal injection/ (kg·h−1 ) Theoretical combustion temperature/ ℃ Fuel rate/ (kg·t−1 ) Oxygen enrichment/ % Sample number 0 200 400 600 800 1000 1200 1400 1600 Sample number 0 200 400 600 800 1000 1200 1400 1600 9000 6000 0.4 0.3 1300 1100 12000 6000 400 300 150 100 50 80 40 600 500 4 2 0 60 2400 2100 30 0 Volume utilization coefficient Synthetic load Gas utilization efficiency/ % Daily hot metal production/t Pressure difference/ MPa Permeability index 2 1 4.0 2.5 55 35 10000 5000 0.25 0.20 0.15 20 10 2×104 1×104 70 60 50 8 4 1520 1440 1.0 0.5 0 0 图 1    铁水温度初选特征参量样本散点图 Fig.1    Scatter plot of the primary characteristic parameters of hot metal temperature · 572 · 工程科学学报,第 43 卷,第 4 期

王振阳等:基于支持向量回归与极限学习机的高炉铁水温度预测 573· 1.0 1.0 (a) Characteristic parameters of positive (b) Characteristic parameters of positive correlation of hot metal temperature correlation of hot metal temperature 0.8 0.8 Characteristic parameters of negative Characteristic parameters of negative correlation of hot metal temperature correlation of hot metal temperature 0.6 0.6 子 0.4 0.4 0.2 0.2 多 0 100-1S 8 Characteristic parameters of hot metal temperature Characteristic parameters of hot metal temperature 图2 铁水温度与初选特征参量之间相关系数.(a)Pearson相关系数:(b)Spearman相关系数 Fig.2 Correlation coefficient between hot metal temperature and primary characteristic parameters:(a)Pearson;(b)Spearman 表2铁水温度终选特征参量 铁水温度预测模型,其中SVR选用高斯RBF核函 Table 2 Final characteristic parameters of hot metal temperature 数,宽度系数为0.01,惩罚系数为10:ELM选用 No. Characteristic parameters Sigmoid激活函数,隐藏节点数设置为100,在此超 1 Fuel rate 参数设置下,两种算法多次预测结果平均值较佳 Nut coke rate SVR与ELM算法的预测结果与测量值的比对如 3 Coke rate 图3所示 Blast temperature 由图3可知,两种算法对铁水温度的追踪结果 5 Bosh gas index 较好,部分样本点预测值与测量值重合度较高.此 6 Permeability index 外,对于与测量值有偏差的预测样本点,其对铁水 7 Hot metal Si content 温度增减变化趋势的预判也较为理想.事实上,对 Daily hot metal production 铁水温度增加或减少的预知将有助于高炉操作者 Current hot metal temperature 预判下一阶段高炉炉缸热状态变化走势,从而可 o Synthetic load 以提前采取应对操作措施,维持高炉稳定顺行,促 Pressure difference 进节能降耗 12 Gas utilization efficiency 图4(a)和(b)分别为基于SVR与ELM两种算 Volume utilization coefficient 法对铁水温度的预测值与测量值所构成的散点 14 Cooling water temperature difference 图.可以看出,两种模型的数据点均围绕对角线方 向延伸分布,且在对角线区域数据点分布较为紧 限学习机模型的训练,使模型不断学习和构建特 密,预测值与测量值吻合度较高.此外,散点主要 征参量与目标参量之间的映射,即将表2所示特 分布在对角线±10℃范围内,说明预测值的偏差 征参量数据与下一日铁水温度数据构建关联映 多集中在±10℃以内.图4(c)为基于SVR与ELM 射,使模型能够实现预测下一阶段铁水温度变化 算法的铁温预测误差概率密度分布函数,经对比 趋势与幅度的目标.而后利用剩余20%的数据构 后发现,相比ELM算法,SVR算法预测误差概率 成测试集,对模型的预测准确度进行验证,并利用 函数分布曲线的延伸范围较小,曲线散点更加集 测试集反馈结果不断优化模型超参数,最终形成 中,说明对测量值的追踪效果较好

限学习机模型的训练,使模型不断学习和构建特 征参量与目标参量之间的映射,即将表 2 所示特 征参量数据与下一日铁水温度数据构建关联映 射,使模型能够实现预测下一阶段铁水温度变化 趋势与幅度的目标. 而后利用剩余 20% 的数据构 成测试集,对模型的预测准确度进行验证,并利用 测试集反馈结果不断优化模型超参数,最终形成 铁水温度预测模型,其中 SVR 选用高斯 RBF 核函 数 ,宽度系数为 0.01,惩罚系数为 10; ELM 选用 Sigmoid 激活函数,隐藏节点数设置为 100,在此超 参数设置下,两种算法多次预测结果平均值较佳. SVR 与 ELM 算法的预测结果与测量值的比对如 图 3 所示. 由图 3 可知,两种算法对铁水温度的追踪结果 较好,部分样本点预测值与测量值重合度较高. 此 外,对于与测量值有偏差的预测样本点,其对铁水 温度增减变化趋势的预判也较为理想. 事实上,对 铁水温度增加或减少的预知将有助于高炉操作者 预判下一阶段高炉炉缸热状态变化走势,从而可 以提前采取应对操作措施,维持高炉稳定顺行,促 进节能降耗. 图 4(a)和(b)分别为基于 SVR 与 ELM 两种算 法对铁水温度的预测值与测量值所构成的散点 图. 可以看出,两种模型的数据点均围绕对角线方 向延伸分布,且在对角线区域数据点分布较为紧 密,预测值与测量值吻合度较高. 此外,散点主要 分布在对角线±10 ℃ 范围内,说明预测值的偏差 多集中在±10 ℃ 以内. 图 4(c)为基于 SVR 与 ELM 算法的铁温预测误差概率密度分布函数,经对比 后发现,相比 ELM 算法,SVR 算法预测误差概率 函数分布曲线的延伸范围较小,曲线散点更加集 中,说明对测量值的追踪效果较好. 表 2    铁水温度终选特征参量 Table 2    Final characteristic parameters of hot metal temperature No. Characteristic parameters 1 Fuel rate 2 Nut coke rate 3 Coke rate 4 Blast temperature 5 Bosh gas index 6 Permeability index 7 Hot metal Si content 8 Daily hot metal production 9 Current hot metal temperature 10 Synthetic load 11 Pressure difference 12 Gas utilization efficiency 13 Volume utilization coefficient 14 Cooling water temperature difference 1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0 (a) (b) Absolute value of correlation coefficient Absolute value of correlation coefficient Characteristic parameters of positive correlation of hot metal temperature Characteristic parameters of negative correlation of hot metal temperature Current hot metal temperature Bosh gas index Gas utilization efficiency Cooling water temperature difference Nut coke rate Hot metal Si-content Fuel rate Volume utilization coefficient Daily hot metal production Synthetic load Coke rate Permeability index Blast temperature Pressure difference Theoretical combustion temperature Pulverized coal injection per hour Blast velocity energy Oxygen enrichment Blast pressure Bosh gas volume Blast volume Coal injection rate Current hot metal temperature Bosh gas index Gas utilization efficiency Cooling water temperature difference Nut coke rate Hot metal Si-content Fuel rate Volume utilization coefficient Daily hot metal production Synthetic load Coke rate Permeability index Blast temperature Pressure difference Theoretical combustion temperature Pulverized coal injection per hour Blast velocity energy Oxygen enrichment Blast pressure Bosh gas volume Blast volume Coal injection rate 0.74 0.40 0.39 0.75 0.40 0.39 0.36 0.33 0.30 0.29 0.28 0.28 0.27 0.24 0.21 0.16 0.15 0.15 0.10 0.10 0.08 0.06 0.05 0.04 0.03 0.35 0.34 0.31 0.31 0.28 0.27 0.25 0.24 0.20 0.12 0.11 0.09 0.08 0.07 0.07 0.05 0.04 0.03 0.03 Characteristic parameters of hot metal temperature Characteristic parameters of hot metal temperature Characteristic parameters of positive correlation of hot metal temperature Characteristic parameters of negative correlation of hot metal temperature 图 2    铁水温度与初选特征参量之间相关系数. (a)Pearson 相关系数;(b)Spearman 相关系数 Fig.2    Correlation coefficient between hot metal temperature and primary characteristic parameters: (a) Pearson; (b) Spearman 王振阳等: 基于支持向量回归与极限学习机的高炉铁水温度预测 · 573 ·

574 工程科学学报,第43卷,第4期 15o0 1500(b 1450 Measured values 1450 values -SVR predictive values Test set sample numbers:1-100 Test set sample numbers:1-100 1500 1500 1450 Measu ed values 1450 Measured values -SVR predictive values Test set sample numbers:101-200 ELM predictive values Test set sample numbers:101-200 1500 1500 1450 Measured values 1450 -SVR predictive values Test set sample numbers:201-300 Test sct sample numbers201-300 1500 8 8 1500 见点安d口 1450 Measured values 1450 Measured values -SVR predictive values Test set sample numbers:301-400 ELM predictive values Test set sample numbers:301-400 20 40 60 80 100 0 20 40 60 80 100 Sample number Sample number 图3铁水温度测量值与预测值比对.(a)基于SVR算法:(b)基于ELM算法 Fig.3 Comparison of measured and predictive values of hot metal temperature:(a)prediction value based on support vector regression(SVR); (b)prediction value based on extreme learning machine(ELM). 图5为SVR和ELM对铁水温度预测结果的 (a) 1510 用日 百分比误差散点分布与统计图.在0.5%范围内 的点个数除上所有点的个数,即为分布概率经统 1490 计计算,SVR和ELM的预测百分比误差在0.5% 1470 以内的分布概率分别为81.0%和78.0%,在±1.0%以 SVR predictive value and 内的分布概率分别为99.0%和983%,说明百分比 measured value comparison 1450 误差主要集中在±0.5%,基本不超过±1.0%.对比两 1450 1470 1490 1510 Measured hot metal temperature/C 种算法预测误差散点分布图,SVR算法更好地控 制了误差偏离严重预测点的数量,因而百分比误 (b) 1510 差点波动较低 0 100 200 300 4000 20 4060 1490 0 1.0. ■SVR percentage error SVR error 0 0.5 1470 g 0 -0.5 -1.0 。g (a) 14 1450 1470 1490 1510 Measured hot metal temperature/C 1.0 ■ELM percentage erro ELMerro 0.5 0 0.08 -0.5 -1.0 0.06 (b) 100 200 300 4000 204060 0.04 图5铁水温度预测的百分比误差散点分布统计图.(a)SVR:(b)ELM 0.02 Fig.5 Scatter distribution statistics of percentage error in hot metal temperature prediction:(a)SVR:(b)ELM -20 -10 0 10 20 对SVR和ELM算法预测结果进行综合分析 Hot metal temperature error/C 算法预测准确度一般应从多方面予以定量化衡量 图4铁水温度预测值与测量值偏差.()基于SVR的铁温预测值与 与表征],本文选取平均绝对误差(Mean absolute 测量值偏差:(b)基于ELM的铁温预测值与测量值偏差:(c)基于 error,,MAE)用以表征模型预测值与测量值之间整 SVR与ELM预测铁温误差概率密度分布函数 体偏差的绝对值;选取平均绝对百分比误差(Mean Fig.4 Deviation of predictive value of hot metal temperature from the absolute percentage error,MAPE)用以表征模型预测 measured value:(a)based on SVR;(b)based on ELM;(c)the probability density distribution function of hot metal temperature error based on SVR 值与测量值之间整体偏差的相对值;选取均方根 and ELM 误差(RMSE,Root mean square error)用以表征模型

图 5 为 SVR 和 ELM 对铁水温度预测结果的 百分比误差散点分布与统计图. 在±0.5% 范围内 的点个数除上所有点的个数,即为分布概率. 经统 计计算,SVR 和 ELM 的预测百分比误差在±0.5% 以内的分布概率分别为 81.0% 和 78.0%,在±1.0% 以 内的分布概率分别为 99.0% 和 98.3%,说明百分比 误差主要集中在±0.5%,基本不超过±1.0%. 对比两 种算法预测误差散点分布图,SVR 算法更好地控 制了误差偏离严重预测点的数量,因而百分比误 差点波动较低. 1.0 0 100 200 300 SVR percentage error SVR error point distribution ELM error point distribution ELM percentage error 400 0 100 200 300 400 0 20 40 60 0 20 40 60 0.5 −0.5 −1.0 0 Percentage error/ % 1.0 0.5 −0.5 −1.0 0 Percentage error/ % (a) (b) 图 5    铁水温度预测的百分比误差散点分布统计图. (a)SVR;(b)ELM Fig.5     Scatter  distribution  statistics  of  percentage  error  in  hot  metal temperature prediction: (a) SVR; (b) ELM 对 SVR 和 ELM 算法预测结果进行综合分析. 算法预测准确度一般应从多方面予以定量化衡量 与表征[25] ,本文选取平均绝对误差(Mean absolute error,MAE)用以表征模型预测值与测量值之间整 体偏差的绝对值;选取平均绝对百分比误差(Mean absolute percentage error,MAPE)用以表征模型预测 值与测量值之间整体偏差的相对值;选取均方根 误差(RMSE,Root mean square error)用以表征模型 Test set sample numbers: 1-100 Test set sample numbers: 1-100 Test set sample numbers: 101-200 Test set sample numbers: 201-300 Test set sample numbers: 301-400 Test set sample numbers: 101-200 Test set sample numbers: 201-300 Test set sample numbers: 301-400 Sample number 1500 (a) (b) 1450 1500 1450 1500 1450 1500 1450 Temperature/ ℃ 1500 1450 1500 1450 1500 1450 1500 1450 Temperature/ ℃ 0 20 40 60 80 100 Sample number 0 20 40 60 80 100 Measured values SVR predictive values Measured values SVR predictive values Measured values SVR predictive values Measured values SVR predictive values Measured values ELM predictive values Measured values ELM predictive values Measured values ELM predictive values Measured values ELM predictive values 图 3    铁水温度测量值与预测值比对. (a)基于 SVR 算法;(b)基于 ELM 算法 Fig.3     Comparison  of  measured  and  predictive  values  of  hot  metal  temperature: (a)  prediction  value  based  on  support  vector  regression  (SVR); (b) prediction value based on extreme learning machine (ELM). Predictive hot metal temperature/ ℃ Measured hot metal temperature/℃ 1510 1510 1490 1490 1470 1470 1450 1450 (a) SVR predictive value and measured value comparison Predictive hot metal temperature/ ℃ Measured hot metal temperature/℃ 1510 1510 1490 1490 1470 1470 1450 1450 (b) ELM predictive value and measured value comparison Probability density of prediction error Hot metal temperature error/℃ 0.08 0.06 0.04 0 0.02 (c) Probability density of SVR prediction error Probability density of ELM prediction error −20 −10 0 10 20 图 4    铁水温度预测值与测量值偏差. (a)基于 SVR 的铁温预测值与 测量值偏差;(b)基于 ELM 的铁温预测值与测量值偏差;(c)基于 SVR 与 ELM 预测铁温误差概率密度分布函数 Fig.4    Deviation of predictive value of hot metal temperature from the measured value: (a) based on SVR; (b) based on ELM; (c) the probability density distribution function of hot metal temperature error based on SVR and ELM · 574 · 工程科学学报,第 43 卷,第 4 期

王振阳等:基于支持向量回归与极限学习机的高炉铁水温度预测 575· 预测值相较测量基准值的偏差波动;选取命中率 及误差波动方面表现较佳,对铁水温度±10℃范 (Hit rate,HR)用以表征在工艺可接受范围内的模 围内的预测命中率为94%. 型预测可信度,其计算方法分别如式(15)至(18) 所示: 参考文献 [1]Wang Z Y,Zhang J L,Liu Z J,et al.Status,technological MAE=-. (15) progress,and development directions of the ironmaking industry in China.Ironmaking Steelmaking,2019,46(10):937 [2] (16) Wang Z Y,Zhang J L,An G,et al.Analysis on the oversize blast furnace desulfurization and a sulfide capacity prediction model based on congregated electron phase.Metall Mater Trans B,2016, RMSE= 日2- (17) 47(1):127 [3]Martin R D,Obeso F,Mochon J,et al.Hot metal temperature prediction in blast furnace using advanced model based on fuzzy R=HR×10% logic tools.Ironmaking Steelmaking,2007,34(3):241 n [4] Jimenez J,Mochon J,de Ayala J S,et al.Blast furnace hot metal (18) 1,-y≤c temperature prediction through neural networks-based models./S// HR;= m,2004.44(3):573 0,-为>c [5] Ding Z Y,Zhang J,Liu Y.Ensemble non-Gaussian local 其中,n为测试集中样本总数量,和:分别为铁 regression for industrial silicon content prediction./SI/Int,2017, 水温度预测值与测量值,c为命中率判定边界值, 57(11):2022 本文选取c值为10℃,即以误差(-10℃,10℃)范 [6] Gao C H.Chen J M.Zeng J S,et al.A chaos-based iterated 围内为基准进行命中率计算,计算结果如表3所 multistep predictor for blast fumace ironmaking process.A/ChE 2009,55(4):947 示.由此可知,无论是从偏差绝对值与相对值角 [7]Cui G M,Li J,Zhang Y,et al.Prediction modeling study for blast 度,亦或是从偏差波动与命中率角度,SVR与ELM fumace hot metal temperature based on T-S fuzzy neural network 算法均对铁水日平均温度实现了较好的预测.相 model.Iron Steel,2013,48(11):11 较而言,SVR算法对铁水温度的预测优于ELM算 (崔桂梅,李静,张勇,等.基于TS模糊神经网络模型的高炉铁 法,其平均绝对误差、平均绝对百分比误差、均方 水温度预测建模.钢铁,2013,48(11):11) 根误差与±10℃以内的命中率表现较佳 [8] Cui G M,Cheng S.Predictive modeling of blast furnace temperature by using distributed neural network model.J Iron 表3SVR与ELM算法铁水温度预测结果综合定量表征 Steel Res,.2014,26(6):27 Table 3 Quantitative characterization of SVR and ELM model (崔桂梅,程史.基于分布式神经网路模型的高炉炉温预测建模 prediction results of hot metal temperature 钢铁研究学报,2014,26(6):27) Model MAPE/% MAE/℃ RMSE/C HP±10℃)M% [9] Shi L,Tang J J,Yu T,et al.Parameters prediction on furnace SVR 0.29 4.33 5.60 94.0 temperature on blast furnace based on a nonlinear spline ELM 0.31 4.69 6.09 88.5 transform-PLS.J Iron Steel Res,2013,25(2):20 (石琳,汤佳佳,于涛,等.基于样条变换的非线性PLS的反应高 炉炉温的参数预测.钢铁研究学报,2013,25(2):20) 3结论 [10]Zhang X M,Kano M,Matsuzaki S.A comparative study of deep and shallow predictive techniques for hot metal temperature (1)正相关影响铁水温度变化的特征参量主 prediction in blast furnace ironmaking.Compu Chem Eng,2019, 要为煤气利用率、铁水硅含量和利用系数等:负相 130:106575 关影响铁水温度变化的特征参量主要为炉腹煤气 [11]Zhang X M,Kano M,Matsuzaki S.Ensemble pattern trees for 指数、水温差和焦丁比等.铁水温度受到高炉直 predicting hot metal temperature in blast furnace.Comput Chem 接与间接还原度以及热量平衡相关特征参量的影 Eng,2019,121:442 [12]Diaz J,Fernandez F J,Prieto MM.Hot metal temperature 响较大 forecasting at steel plant using multivariate adaptive regression (2)基于SVR与ELM算法构建的模型均对铁 splines.Metals,2020,10(1):41 水温度实现了较好的预测效果,前者在预测值与 [13]Hashimoto Y,Sawa Y,Kano M.Online prediction of hot metal 误差值散点分布,平均绝对误差和百分比误差,以 temperature using transient model and moving horizon estimation

预测值相较测量基准值的偏差波动;选取命中率 (Hit rate,HR)用以表征在工艺可接受范围内的模 型预测可信度,其计算方法分别如式(15)至(18) 所示: MAE = 1 n · ∑n i=1 |yˆi −yi | (15) MAPE = 1 n · ∑n i=1 yˆi −yi yi ×100% (16) RMSE = tv 1 n · ∑n i=1 (yˆi −yi) 2 (17)    HR = 1 n · ∑n i=1 HRi ×100% HRi =    1, byi −yi ⩽ c 0, byi −yi > c (18) 其中,n 为测试集中样本总数量, yˆi 和 yi 分别为铁 水温度预测值与测量值,c 为命中率判定边界值, 本文选取 c 值为 10 ℃,即以误差(−10 ℃,10 ℃)范 围内为基准进行命中率计算,计算结果如表 3 所 示. 由此可知,无论是从偏差绝对值与相对值角 度,亦或是从偏差波动与命中率角度,SVR 与 ELM 算法均对铁水日平均温度实现了较好的预测. 相 较而言,SVR 算法对铁水温度的预测优于 ELM 算 法,其平均绝对误差、平均绝对百分比误差、均方 根误差与±10 ℃ 以内的命中率表现较佳. 表 3  SVR 与 ELM 算法铁水温度预测结果综合定量表征 Table  3    Quantitative  characterization  of  SVR  and  ELM  model prediction results of hot metal temperature Model MAPE/% MAE /℃ RMSE/℃ HP(±10 ℃)/% SVR 0.29 4.33 5.60 94.0 ELM 0.31 4.69 6.09 88.5 3    结论 (1)正相关影响铁水温度变化的特征参量主 要为煤气利用率、铁水硅含量和利用系数等;负相 关影响铁水温度变化的特征参量主要为炉腹煤气 指数、水温差和焦丁比等. 铁水温度受到高炉直 接与间接还原度以及热量平衡相关特征参量的影 响较大. (2)基于 SVR 与 ELM 算法构建的模型均对铁 水温度实现了较好的预测效果,前者在预测值与 误差值散点分布,平均绝对误差和百分比误差,以 及误差波动方面表现较佳,对铁水温度±10 ℃ 范 围内的预测命中率为 94%. 参    考    文    献 Wang  Z  Y,  Zhang  J  L,  Liu  Z  J,  et  al.  Status,  technological progress, and development directions of the ironmaking industry in China. Ironmaking Steelmaking, 2019, 46(10): 937 [1] Wang Z Y, Zhang J L, An G, et al. Analysis on the oversize blast furnace  desulfurization  and  a  sulfide  capacity  prediction  model based on congregated electron phase. Metall Mater Trans B, 2016, 47(1): 127 [2] Martin  R  D,  Obeso  F,  Mochon  J,  et  al.  Hot  metal  temperature prediction in blast furnace using advanced model based on fuzzy logic tools. Ironmaking Steelmaking, 2007, 34(3): 241 [3] Jimenez J, Mochon J, de Ayala J S, et al. Blast furnace hot metal temperature prediction through neural networks-based models. ISIJ Int, 2004, 44(3): 573 [4] Ding  Z  Y,  Zhang  J,  Liu  Y.  Ensemble  non-Gaussian  local regression for industrial silicon content prediction. ISIJ Int, 2017, 57(11): 2022 [5] Gao  C  H,  Chen  J  M,  Zeng  J  S,  et  al.  A  chaos-based  iterated multistep predictor for blast furnace ironmaking process. AIChE J, 2009, 55(4): 947 [6] Cui G M, Li J, Zhang Y, et al. Prediction modeling study for blast furnace hot metal temperature based on T-S fuzzy neural network model. Iron Steel, 2013, 48(11): 11 (崔桂梅, 李静, 张勇, 等. 基于T-S模糊神经网络模型的高炉铁 水温度预测建模. 钢铁, 2013, 48(11):11) [7] Cui  G  M,  Cheng  S.  Predictive  modeling  of  blast  furnace temperature  by  using  distributed  neural  network  model. J Iron Steel Res, 2014, 26(6): 27 (崔桂梅, 程史. 基于分布式神经网络模型的高炉炉温预测建模. 钢铁研究学报, 2014, 26(6):27) [8] Shi  L,  Tang  J  J,  Yu  T,  et  al.  Parameters  prediction  on  furnace temperature  on  blast  furnace  based  on  a  nonlinear  spline transform-PLS. J Iron Steel Res, 2013, 25(2): 20 (石琳, 汤佳佳, 于涛, 等. 基于样条变换的非线性PLS的反应高 炉炉温的参数预测. 钢铁研究学报, 2013, 25(2):20) [9] Zhang X M, Kano M, Matsuzaki S. A comparative study of deep and  shallow  predictive  techniques  for  hot  metal  temperature prediction in blast furnace ironmaking. Comput Chem Eng, 2019, 130: 106575 [10] Zhang  X  M,  Kano  M,  Matsuzaki  S.  Ensemble  pattern  trees  for predicting  hot  metal  temperature  in  blast  furnace. Comput Chem Eng, 2019, 121: 442 [11] Diaz  J,  Fernandez  F  J,  Prieto  M  M.  Hot  metal  temperature forecasting  at  steel  plant  using  multivariate  adaptive  regression splines. Metals, 2020, 10(1): 41 [12] Hashimoto  Y,  Sawa  Y,  Kano  M.  Online  prediction  of  hot  metal temperature using transient model and moving horizon estimation. [13] 王振阳等: 基于支持向量回归与极限学习机的高炉铁水温度预测 · 575 ·

576 工程科学学报,第43卷,第4期 ISUm,2019,59(9):1534 39 [14]Su X L,Zhang S,Yin Y X,et al.Prediction model of hot metal (张海刚,张森,尹怡欣.基于全局优化支持向量机的多类别高 temperature for blast furnace based on improved multi-layer 炉故障诊断.工程科学学报,2017,39(1):39) extreme leaming machine.Int J Mach Learn Cybern,2019, [21]Chu F,Ye J F,Ma X P,et al.Online performance prediction of 10(10):2739 CCPP byproduct coal-gas system based on online sequential [15]Zhao H,Zhao D T,Yue Y J,et al.Study on prediction method of extreme learning machine.ChinJ Eng,2016,38(6):861 hot metal temperature in blast fumace IEEE International (褚菲,叶俊锋,马小平,等.基于OS-ELM的CCPP副产煤气燃料 Conference on Mechatronics and Automation.Takamatsu,2017: 系统在线性能预测.工程科学学报,2016,38(6):861) 316 [22]Li A L,Zhao Y M,Cui G M.Prediction model of blast furnace [16]Wang Y K,Liu X Y,Zhang B L.On feature selection and blast temperature based on ELM with grey correlation analysis.J Iron fumace temperature tendency prediction in hot metal based on Steel Res,.2015,27(11):33 SVM-RFE II Australian and New Zealand Control Conference (李爱莲,赵永明,崔桂梅.基于灰色关联分析的ELM高炉温度 Melbourne,2018:371 预测模型.钢铁研究学报,2015,27(11):33) [17]Yue Y J.Dong A.Zhao H,et al.Study on prediction model of [23]Chen H Z,Yang J P,Lu X C,et al.Quality prediction of the blast fumace hot metal temperature /IEEE International continuous casting bloom based on the extreme learning machine. Conference on Mechatronics and Automation.Harbin,2016:1396 Chin J Eng,2018,40(7):815 [18]Liu Y,Gao Z.Enhanced just-in-time modelling for online quality (陈恒志,杨建平,卢新春,等.基于极限学习机(ELM)的连铸坯 prediction in BF ironmaking.Ironmaking Steelmaking,2015, 质量预测.工程科学学报,2018,40(7):815) 42(5):321 [24]Chen P,Li Q,Zhang D Z,et al.A survey of multimodal machine [19]Guo D W,Zhou P.Soft-sensor modeling of silicon content in hot learning.Chin J Eng,2020,42(5):557 metal based on sparse robust LS-SVR and multi-objective (陈鹏,李擎,张德政,等.多模态学习方法综述.工程科学学报, optimization.Chin J Eng,2016,38(9):1233 2020,42(5):557) (郭东伟,周平,基于稀疏化鲁棒LS-SVR与多目标优化的铁水硅 [25]Tunckaya Y.Performance assessment of permeability index 含量软测量建模.工程科学学报,2016,38(9):1233) prediction in an ironmaking process via soft computing techniques. [20]Zhang H G,Zhang S,Yin Y X.Multi-class fault diagnosis of BF Proc Inst Mech Eng Part E-J Process Mech Eng,2017,231(6): based on global optimization LS-SVM.Chin J Eng,2017,39(1) 1101

ISIJ Int, 2019, 59(9): 1534 Su X L, Zhang S, Yin Y X, et al. Prediction model of hot metal temperature  for  blast  furnace  based  on  improved  multi-layer extreme  learning  machine. Int J Mach Learn Cybern,  2019, 10(10): 2739 [14] Zhao H, Zhao D T, Yue Y J, et al. Study on prediction method of hot  metal  temperature  in  blast  furnace  // IEEE International Conference on Mechatronics and Automation. Takamatsu,  2017: 316 [15] Wang Y K, Liu X Y, Zhang B L. On feature selection and blast furnace  temperature  tendency  prediction  in  hot  metal  based  on SVM-RFE  // Australian and New Zealand Control Conference. Melbourne, 2018: 371 [16] Yue  Y  J,  Dong  A,  Zhao  H,  et  al.  Study  on  prediction  model  of blast  furnace  hot  metal  temperature  // IEEE International Conference on Mechatronics and Automation. Harbin, 2016: 1396 [17] Liu Y, Gao Z. Enhanced just-in-time modelling for online quality prediction  in  BF  ironmaking. Ironmaking Steelmaking,  2015, 42(5): 321 [18] Guo D W, Zhou P. Soft-sensor modeling of silicon content in hot metal  based  on  sparse  robust  LS-SVR  and  multi-objective optimization. Chin J Eng, 2016, 38(9): 1233 (郭东伟, 周平. 基于稀疏化鲁棒LS-SVR与多目标优化的铁水硅 含量软测量建模. 工程科学学报, 2016, 38(9):1233) [19] Zhang H G, Zhang S, Yin Y X. Multi-class fault diagnosis of BF based on global optimization LS-SVM. Chin J Eng, 2017, 39(1): [20] 39 (张海刚, 张森, 尹怡欣. 基于全局优化支持向量机的多类别高 炉故障诊断. 工程科学学报, 2017, 39(1):39) Chu  F,  Ye  J  F,  Ma  X  P,  et  al.  Online  performance  prediction  of CCPP  byproduct  coal-gas  system  based  on  online  sequential extreme learning machine. Chin J Eng, 2016, 38(6): 861 (褚菲, 叶俊锋, 马小平, 等. 基于OS-ELM的CCPP副产煤气燃料 系统在线性能预测. 工程科学学报, 2016, 38(6):861) [21] Li  A  L,  Zhao  Y  M,  Cui  G  M.  Prediction  model  of  blast  furnace temperature based on ELM with grey correlation analysis. J Iron Steel Res, 2015, 27(11): 33 (李爱莲, 赵永明, 崔桂梅. 基于灰色关联分析的ELM高炉温度 预测模型. 钢铁研究学报, 2015, 27(11):33) [22] Chen  H  Z,  Yang  J  P,  Lu  X  C,  et  al.  Quality  prediction  of  the continuous casting bloom based on the extreme learning machine. Chin J Eng, 2018, 40(7): 815 (陈恒志, 杨建平, 卢新春, 等. 基于极限学习机(ELM)的连铸坯 质量预测. 工程科学学报, 2018, 40(7):815) [23] Chen P, Li Q, Zhang D Z, et al. A survey of multimodal machine learning. Chin J Eng, 2020, 42(5): 557 (陈鹏, 李擎, 张德政, 等. 多模态学习方法综述. 工程科学学报, 2020, 42(5):557) [24] Tunckaya  Y.  Performance  assessment  of  permeability  index prediction in an ironmaking process via soft computing techniques. Proc Inst Mech Eng Part E-J Process Mech Eng,  2017,  231(6): 1101 [25] · 576 · 工程科学学报,第 43 卷,第 4 期

点击下载完整版文档(PDF)VIP每日下载上限内不扣除下载券和下载次数;
按次数下载不扣除下载券;
24小时内重复下载只扣除一次;
顺序:VIP每日次数-->可用次数-->下载券;
已到末页,全文结束
相关文档

关于我们|帮助中心|下载说明|相关软件|意见反馈|联系我们

Copyright © 2008-现在 cucdc.com 高等教育资讯网 版权所有