作者：崔桂梅, 张运强, 张勇

作者单位：内蒙古科技大学信息工程学院, 内蒙古 包头 014010

关键词：回旋区温度预测;核极限学习机;粒子群优化;高炉

摘要：

高炉平稳运行时，喷煤量常根据回旋区温度的波动随时进行调整。相较于人工推断回旋区温度，建立回旋区温度预测模型更能及时准确地对喷煤量进行优化。针对常规PSO-KELM算法对回旋区温度预测命中率较低的问题，提出改进的PSO-KELM算法对其进行建模预测。首先，采用混沌机制调整惯性权重，并线性改变学习因子，以平衡粒子群的全局与局部搜索能力。其次，针对粒子群算法易陷入局部最优的缺点，引入遗传算法思想，将种群粒子交叉、变异，以提高种群多样性。最后，基于所提方法用某高炉运行数据建立回旋区温度预测模型，并与用常规PSO-KELM算法、BP神经网络、极限学习机建立的模型作对比。仿真结果表明，用所提算法建立的模型预测回旋区温度具有最高命中率和最低均方误差。

Prediction of the raceway temperature based on improved PSO-KELM
CUI Guimei, ZHANG Yunqiang, ZHANG Yong
School of Information Engineering, Inner Mongolia University of Science and Technology, Baotou 014010, China
Abstract: When the blast furnace runs in a stable way, the coal injection quantity is frequently adjusted on the basis of temperature fluctuation of raceway in blast furnace. Compared with deducing artificially, the coal injection can be optimized timely and precisely by establishing a raceway temperature prediction model. To improve the hitting accuracy of the traditional PSO-KELM algorithm, an improved PSO-KELM algorithm was proposed. In the first place, chaotic mechanism was adopted to adjust inertia weight and alter learning factors linearly in order that the global and local searching ability can be balanced. In addition, the genetic algorithm, which can solve the problem of local optimum in PSO was introduced to promote the diversity of population by crossing and mutating population particles. Finally, compared with traditional PSO-KELM algorithm, BP neural network and extreme learning machine, the simulation results demonstrated that the raceway temperature prediction model, which was established based upon the new method and operating data of a blast furnace, was able to predict raceway temperature with the maximum hitting accuracy and the minimum mean square error.
Keywords: prediction of raceway temperature;kernel extreme learning machine;particle swarm optimization;blast furnace
2020, 46(4):25-30  收稿日期: 2019-10-16;收到修改稿日期: 2019-11-15
基金项目: 国家自然科学基金项目（61763039）
作者简介: 崔桂梅(1963-),女,河北保定市人,教授,博士,主要研究方向为复杂过程系统的建模及运行优化控制
参考文献
[1] 崔大福, 刘先龙, 江杰,等. 高炉风口回旋区测温成像人工窥视在线监测系统的研究与应用[C]//中国计量协会冶金分会2018年会论文集, 2018:32-34.
[2] 温良英, 欧阳奇, 白晨光, 等. 基于机器视觉的风口回旋区温度检测算法[J]. 钢铁研究学报, 2007, 19(2):5-9
[3] 李严, 程树森. 基于蒙特卡洛法的高炉回旋区温度测量研究[C]//2017年全国高炉炼铁学术年会论文集(下), 2017:262-269.
[4] 王洪桥, 程树森, 赵立英, 等. 基于CCD的高炉风口回旋区三维温度场的检测技术[J]. 工程科学学报, 2015, 37(8):1069-1075
[5] 金焱, 袁辉, 祝俊俊, 等. 高炉风口回旋区内温度分布的模拟分析[J]. 武汉科技大学学报, 2017, 40(1):1-4
[6] HUANG G B, ZHOU H M, DING X J, et al. Extreme learning machine for regression and multiclass classification[J]. IEEE Trans on Systems, Man, and Cybernetics, Part B:Cybernetics, 2012, 42(2):513-529
[7] 盛晓晨, 史旭东, 熊伟丽. 改进粒子群优化的极限学习机软测量建模方法[J]. 计算机应用研究, 2019, 37(6):1-6
[8] KENNEDY J E. Particle swarm optimization[C]//Proceedings of the IEEE International Conference on Neural Networks, Perth, Australia, 1995:1942-1948.
[9] 余胜威. MATLAB优化算法案例分析与应用[M]. 北京:清华大学出版社, 2014:203-207.
[10] FENG Y, TENG G F, WANG A X, et al. Chaotic inertia weight in particle swarm optimization[C]//ICICIC 2007:Proceedings of the Second International Conference on Innovative Computing, Information and Control. Piscataway, NJ:IEEE, 2007:475-476.
[11] BANSAL J C, SINGH P K, SARAWAT M, et al. Inertia weight strategies in particle swarm optimization[C]//Proceedings of the 3rd World Congress on Nature and Biologically Inspired Computing(NaBIC), 2011:633-640.
[12] 蒋炜, 彭新一, 周育人. 基于改进PSO的基因调控网络重构方法[J]. 计算机工程, 2009, 35(20):181-183
[13] 施兢业, 刘俊, 陆龚琪. 电力设备红外图像的小波阈值去噪方法研究[J]. 光学技术, 2017, 43(2):162-165
[14] GAO J, WANG Q. BM3D image denoising algorithm based on K-Means clustering[C]//China Academic Conference on Printing & Packaging and Media Technology. Springer, Singapore, 2016:265.
[15] 孙元, 彭小奇, 唐英. 基于彩色CCD的比色测温校正方法[J]. 仪器仪表学报, 2008, 29(1):49-54