作者：李海军¹, 夏静¹, 史恒惠¹, 刘长良^2,3, 王梓齐³

作者单位：1. 国家电投集团河南电力有限公司技术信息中心, 河南 郑州 450001;
2. 华北电力大学 新能源电力系统国家重点实验室, 北京 102206;
3. 华北电力大学控制与计算机工程学院, 河北 保定 071000

关键词：复相关系数;时滞联合估计;NO_x排放;软测量

摘要：

针对工业过程的软测量建模，为对输入与输出变量间的时滞关系进行准确、快速地估计，提出一种基于复相关系数的时滞联合估计方法。该方法以模型输入和输出数据间的复相关系数为指标，将时滞联合估计问题转化为多维优化问题，进而对各输入变量的时滞时间进行寻优。针对火电厂脱硝系统的NO_x排放软测量，基于实际的运行数据和最小二乘支持向量机算法，对所提出的方法进行验证并与其他时滞估计方法进行对比。结果表明：基于复相关系数的时滞估计方法计算速度较快，时滞估计结果较准确，能在一定程度上提高软测量模型的准确度。

Time-delay joint estimation based on multiple correlation coefficient and its application
LI Haijun¹, XIA Jing¹, SHI Henghui¹, LIU Changliang^2,3, WANG Ziqi³
1. Technical Information Center of Henan Electric Power Co., Ltd. of State Power Investment Group, Zhengzhou 450001, China;
2. State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power University, Beijing 102206, China;
3. School of Control and Computer Engineering, North China Electric Power University, Baoding 071000, China
Abstract: In order to accurately and quickly estimate the time-delay relationship between input and output variables, a joint time-delay estimation method based on multiple correlation coefficient is proposed for the soft sensor modeling of industrial processes. This method takes the multiple correlation coefficients between the input and output data of the model as the index, transforms the time-delay joint estimation problem into a multi-dimensional optimization problem and then optimizes the time-delay of each input variable. Based on the practical operation data and the least square support vector machine, the proposed method was verified and compared with other time-delay estimation methods for the soft sensor of NO_x emission in the thermal power plant denitrification system. The results show that the proposed multiple correlation coefficient method is faster and more accurate, and the accuracy of soft sensor model can be improved to a certain extent.
Keywords: multiple correlation coefficient;time-delay joint estimation;NO_x emission;soft sensor
2019, 45(8):140-144  收稿日期: 2018-07-16;收到修改稿日期: 2018-09-04
基金项目: 北京市自然科学基金资助项目（4182061）
作者简介: 李海军(1973-),男,河南焦作市人,工程师,研究方向为火电机组软测量、远程诊断
参考文献
[1] XIONG W, LI Y, ZHAO Y, et al. Adaptive soft sensor based on time difference Gaussian process regression with local time-delay reconstruction[J]. Chemical Engineering Research&Design, 2017, 117:670-680
[2] 申贵成,管水城,孙方宇.考虑最优时滞因子时空模型的高速公路短时交通流预测[J].科学技术与工程, 2018, 18(24):149-156
[3] LOU H, SU H, XIE L, et al. Inferential model for industrial polypropylene melt index prediction with embedded priori knowledge and delay estimation[J]. Industrial&Engineering Chemistry Research, 2012, 51(25):8510-8525
[4] 李妍君,熊伟丽,徐保国.一种带过程变量时滞估计的在线软测量建模方法[J].信息与控制, 2016, 45(6):641-646
[5] 熊伟丽,李妍君.选择性集成LTDGPR模型的自适应软测量建模方法[J].化工学报, 2017, 68(3):984-991
[6] 阮宏镁,田学民,王平.基于联合互信息的动态软测量方法[J].化工学报, 2014, 65(11):4497-4502
[7] 王钧炎,黄德先.基于混合差分进化算法的软测量时延参数估计[J].化工学报, 2008, 59(8):2058-2064
[8] 阮宏镁,田学民,王平.带时延估计的时间差分PLS软测量建模方法[J].石油化工自动化, 2013, 49(6):35-39
[9] MONDINI V, MANGIA A L, TALEVI L, et al. Sinc-windowing and multiple correlation coefficients improve SSVEP recognition based on canonical correlation analysis[J]. Computational Intelligence and Neuroscience, 2018, 2018:4278782
[10] 管燕,吴和成,黄舜.基于改进DEA的江苏省科技资源配置效率研究[J].科研管理, 2011, 32(2):145-150
[11] 范德成,李昊,刘贇.基于改进DEA——以复相关系数为基准的滞后期的我国产业结构演化效率评价[J].运筹与管理, 2016, 25(3):195-203
[12] 秦天牧,吕游,杨婷婷,等. SCR烟气脱硝系统自适应混合动态模型[J].仪器仪表学报, 2016, 37(12):2844-2850
[13] 马平,李珍,梁薇.基于互信息的辅助变量筛选及在火电厂NO_x软测量模型中的应用[J].科学技术与工程, 2017, 17(22):249-254
[14] 丁续达,刘潇,金秀章.基于压缩感知最小二乘支持向量机的NOx软测量模型[J].热力发电, 2018, 47(3):76-81