复杂化工过程失配子模型深度诊断与修正算法

doi:10.11949/j.issn.0438-1157.20180907

摘要/Abstract

摘要：

针对复杂化工生产过程中多变量预测控制系统发生模型失配导致控制性能下降造成的产品质量波动的问题，研究了过程失配子模型深度诊断与模型修正方法。考虑到复杂化工生产过程中被控变量为控制通道和扰动通道的综合响应，通过逐次移动操作变量到扰动通道的方法，评价移动后对模型质量指标的影响，从而判断出所移出子模型的性能，进而对失配子模型进行定位。进一步地，利用现场采集的历史数据用自回归滑动平均模型辨识法辨识出失配部分的模型，用于对原有子模型进行修正。实验采用Wood-Berry精馏过程对其进行动态仿真验证，结果证明了该算法的有效性。

关键词: 复杂化工过程, 模型预测控制, 动力学模型, 修正算法, 动态仿真

Abstract:

In the processes of complicated chemical production, the appearance of model mismatch in multivariable predictive control systems usually will cause fluctuations in product quality. To deal with this problem, a deep dive mismatched sub-models diagnostic method and its correction algorithm are proposed. Considering that the control variable is the combined responses of the control channel and the disturbance channel in industrial field, by the successive moving of the operational variables into disturbance channel, we evaluate the impact of the moving on the quality index of the model, then identify the source of the mismatched sub-model according to its performance judging. Furthermore, the historical data were used to identify the details of the mismatched parts of the model by the autoregressive moving average model identification method, so that the original sub-model can be corrected. The dynamic simulation of the experiment was carried out by Wood-Berry distillation process. The results demonstrate the effectiveness of the algorithm.

Key words: complicated chemical processes, model-predictive control, kinetic modeling, correction algorithm, dynamic simulation

中图分类号:

TP202

杨世品, 黄振, 李丽娟, 宋健全, 叶景, 汪辉. 复杂化工过程失配子模型深度诊断与修正算法[J]. 化工学报, 2019, 70(4): 1485-1493.

Shipin YANG, Zhen HUANG, Lijuan LI, Jianquan SONG, Jing YE, Hui WANG. A deep dive diagnostic and correction algorithm for mismatched sub-models in complicated chemical processes[J]. CIESC Journal, 2019, 70(4): 1485-1493.

图/表 10

图1 MIMO预测控制系统原理

Fig.1 Schematic diagram of MIMO predictive control system

表1 Wood-Berry精馏塔单模型失配仿真测试结果

Table 1 Simulation test results of Wood-Berry distillation column with single model mismatch

序号	控制模型	MQI1	MQI2
Case1 无失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9355	0.9862
Case2 子模型失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1$	0.9285	0.4402
Case3 移除CV₂-MV₁通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1$	0.9221	0.3572
Case4 移除CV₂-MV₂通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1$	0.9237	0.5695
模型修正	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1 + Δ G 22$	0.9347	0.9262

表1 Wood-Berry精馏塔单模型失配仿真测试结果

Table 1 Simulation test results of Wood-Berry distillation column with single model mismatch

序号	控制模型	MQI1	MQI2
Case1 无失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9355	0.9862
Case2 子模型失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1$	0.9285	0.4402
Case3 移除CV₂-MV₁通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1$	0.9221	0.3572
Case4 移除CV₂-MV₂通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1$	0.9237	0.5695
模型修正	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 1.8 × q - 4 - 1.3472 1 - 0.65 q - 1 + Δ G 22$	0.9347	0.9262

图2 模型精确时y 1和y 2的实际浓度曲线

Fig.2 Actual concentration curve of y 1 and y 2 with accurate model

图3 模型失配时y 1和y 2的实际浓度曲线

Fig.3 Actual concentration curve of y 1 and y 2 with mismatched model

图4 模型修正后y 1和y 2的实际浓度曲线

Fig.4 Actual concentration curve of y 1 and y 2 after the mismatched model was corrected

图5 整个失配-修正过程的y 2实际输出绝对误差

Fig.5 y 2 absolute error during the whole mismatching and correcting process

表2 Wood-Berry精馏塔多模型失配仿真测试结果

Table 2 Simulation test results of Wood-Berry distillation column with multiple model mismatch

序号	控制模型	MQI1	MQI2
Case1 无失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9759	0.9298
Case2 子模型失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9646	0.2856
Case3 移除CV₂-MV₁通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9601	0.3532	κ ₂₍₁₎>1 1.2367
Case4 移除CV₂-MV₂通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9541	0.3275	κ ₂₍₂₎>1 1.1467
模型修正	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 + Δ G 21 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1 + Δ G 22$	0.9596	0.9233

表2 Wood-Berry精馏塔多模型失配仿真测试结果

Table 2 Simulation test results of Wood-Berry distillation column with multiple model mismatch

序号	控制模型	MQI1	MQI2
Case1 无失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 q - 8 0.6055 1 - 0.9123 q - 3 q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9759	0.9298
Case2 子模型失配	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9646	0.2856
Case3 移除CV₂-MV₁通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9601	0.3532	κ ₂₍₁₎>1 1.2367
Case4 移除CV₂-MV₂通道	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1$	0.9541	0.3275	κ ₂₍₂₎>1 1.1467
模型修正	$q - 2 0.7665 1 - 0.9419 q - 1 q - 4 - 0.9000 1 - 0.9535 q - 1 3 × q - 8 0.5 1 - 0.75 q - 3 + Δ G 21 1.5 × q - 4 - 1.3472 1 - 0.9329 q - 1 + Δ G 22$	0.9596	0.9233

图6 两个模型失配时y 1和y 2的实际浓度曲线

Fig.6 Actual concentration curve of y 1 and y 2 with double mismatched models

图7 两个模型修正后y 1和y 2的实际浓度曲线

Fig.7 Actual concentration curve of y 1 and y 2 after double mismatched models were corrected

图8 整个失配-修正过程的y 2实际输出绝对误差(多模型失配)

Fig.8 y 2 absolute error during the whole mismatching and correcting process(multiple model mismatch)

参考文献 30

1	Badwe A S , Patwardhan R S , Shah S L , et al . Quantifying the impact of model-plant mismatch on controller performance[J]. Journal of Process Control, 2010, 20(4): 408-425.
2	王晓丽, 黄蕾, 杨鹏, 等 . 动态RBF神经网络在浮选过程模型失配中的应用[J]. 化工学报, 2016, 67(3): 897-902.
	Wang X L , Huang L , Yang P , et al . Dynamic RBF neural networks for model mismatch problem and its application in flotation process [J]. CIESC Journal , 2016, 67(3): 897-902.
3	王丽辉, 汤健彬, 余世明 . 克服广义预测控制模型参数失配的PID反馈校正策略[J]. 浙江工业大学学报, 2005, 33(3): 265-267.
	Wang L H , Tang J B , Yu S M . PID feedback correction strategy overcoming model parametric mismatch of GPC[J]. Journal of Zhejiang University of Technology, 2005, 33(3): 265-267.
4	陈贵, 杨江, 谢磊, 等 . 基于子空间方法的模型失配检测研究[J]. 化工学报, 2011, 62(9): 2575-2581.
	Chen G , Yang J , Xie L , et al . Research on model-plant mismatch detection based on subspace approach[J]. CIESC Journal , 2011, 62(9): 2575-2581.
5	Wang H , Xie L , Song Z H . A review for model plant mismatch measures in process monitoring[J]. Chinese Journal of Chemical Engineering, 2012, 20(6): 1039-1046.
6	Yin F , Wang H , Xie L , et al . Data driven model mismatch detection based on statistical band of Markov parameters [J]. Computers Electrical Engineering, 2014, 40(7): 2178-2192.
7	李秋美, 田学民, 尚林源 . 基于偏相关性分析的MPC控制器模型失配检测[J]. 化工学报, 2016, 67(3): 852-857.
	Li Q M , Tian X M , Shang L Y . Detection of model-plant mismatch based on partial correlation analysis of MPC controllers[J]. CIESC Journal, 2016, 67(3): 852-857.
8	郑恩让, 阮士涛, 王黎 . 基于失配模型的内模控制器的设计[J]. 控制工程, 2016, 23(4): 532-537.
	Zheng E R , Ruan S T , Wang L . Design of internal model controller based on mismatched model[J]. Control Engineering of China, 2016, 23(4): 532-537.
9	张晓晓, 李丽娟, 宋健全, 等 . 预测控制模型失配的系统诊断方法[J]. 计算机工程与设计, 2017, 38(10): 2729-2734.
	Zhang X X , Li L J , Song J Q , et al . System diagnosis algorithm for predictive control model mismatch[J]. Computer Engineering and Design, 2017, 38(10): 2729-2734.
10	Mandur J S , Budman H M . Simultaneous model identification and optimization in presence of model-plant mismatch[J]. Chemical Engineering Journal, 2015, 129: 106-115.
11	宋莎莎, 赵忠盖, 刘飞 . 模型参数失配有界下的扩展集员估计方法[J]. 控制理论与应用, 2017, 34(5): 648-654.
	Song S S , Zhao Z G , Liu F . Extended set membership filtering method for process with bounded-mismatch parameters[J]. Control Theory & Applications, 2017, 34(5): 648-654.
12	Webber J R , Gupta Y P . A closed-loop cross-correlation method for detecting model mismatch in MIMO model-based controllers[J]. ISA Transactions, 2008, 47(4): 395-400.
13	Ma H , Yang M , Hu J . A method of control performance diagnosis to discriminate between model-plant mismatch and unmeasured disturbance[C]//Control and Decision Conference, 2011: 2930-2934.
14	Rengaswamy R , Hägglund T , Venkatasubramanian V . A qualitative shape analysis formalism for monitoring control loop performance [J]. Engineering Applications of Artificial Intelligence, 2001, 14(1): 23-33.
15	王宇红, 刘华怡, 王学剑 . 基于互相关分析的模型失配性能评价与监视[J]. 控制工程, 2011, 18(6): 905-909.
	Wang Y H , Liu H Y , Wang X J . Performance assessment and monitoring of model–plant mismatch based on cross-correlation[J]. Control Engineering of China, 2011, 18(6): 905-909.
16	Carlsson R . A practical approach to detection of plant model mismatch for MPC[D]. Reglerteknik: Linköpings University, 2010.
17	Conner J S , Seborg D E . Assessing the need for process reidentification[J]. Industrial & Engineering Chemistry Research, 2005, 44(8): 2767-2775.
18	Selvanathan S , Tangirala A K . Diagnosis of poor control loop performance due to model-plant mismatch [J]. Industrial & Engineering Chemistry Research, 2010, 49(9): 4210-4229.
19	Ji G , Zhang K , Zhu Y . A method of MPC model error detection[J]. Journal of Process Control, 2012, 22(3): 635-642.
20	Pannocchia G , De Luca A . Performance degradation diagnosis and remedies in offset-free MPC[C]// Proceedings of the American Control Conference, 2012: 429-434.
21	吴夏来, 楼赣菲, 陈超, 等 . 一种克服模型参数失配的广义预测控制算法[J]. 机电工程, 2012, 29 (10): 1232-1234.
	Wu X L , Lou G F , Chen C , et al . Generalized predictive control algorithm overcoming model parametric mismatch [J]. Journal of Mechanical & Electrical Engineering, 2012, 29(10): 1232-1234.
22	魏桂芬, 赵曜 . 基于选择性自校正的模糊动态矩阵控制[J]. 控制工程, 2007, 14(3): 304-306.
	Wei G F , Zhao Y . Fuzzy dynamic matrix control based on selective self-tuning[J]. Control Engineering of China, 2007, 14(3): 304-306.
23	Shardt Y A W , Huang B . Closed-loop identification with routine operating data: effect of time delay and sampling time [J]. Journal of Process Control, 2011, 21(7): 997-1010.
24	吴上, 孙晓, 胡威林, 等 . 基于T-S模糊模型的供水系统预测控制策略研究[J]. 湖南工业大学学报, 2015, 29(4): 85-89.
	Wu S , Sun X , Hu W L , et al . Predictive control of water supply system based on T-S fuzzy model[J]. Journal of Hunan University of Technology, 2015, 29(4): 85-89.
25	Ljung L . System Identification: Theory for the User [M]. Upper Saddle River, NJ: Prentice Hall PRT , 1999.
26	Qin S J . An overview of subspace identification[J]. Computers & Chemical Engineering, 2006, 30(10/11/12): 1502-1513.
27	Hannan J , Dunsmuir W T M , Deistler M . Estimation of vector ARMAX models[J]. Journal of Multivariate Analysis, 1980, 10(3): 275-295.
28	Reddy D C , Deergha R K . An adaptive algorithm for nonlinear system identification[J]. Sadhana, 1991, 16(3): 263-274.
29	Wood R K , Berry M W . Terminal composition control of a binary distillation colomn[J]. Chemical Engineering Science, 1973, 28(9): 1707-1717.
30	王宇红, 王学剑 . 基于协方差基准的模型预测控制性能评价与监视[J]. 信息与控制, 2010, 39(6): 694-699.
	Wang Y H , Wang X J . Performance assessment and monitoring of model predictive control based on covariance benchmark[J]. Information and Control, 2010, 39(6): 694-699.

[1]	杨欣, 王文, 徐凯, 马凡华. 高压氢气加注过程中温度特征仿真分析[J]. 化工学报, 2023, 74(S1): 280-286.
[2]	宋嘉豪, 王文. 斯特林发动机与高温热管耦合运行特性研究[J]. 化工学报, 2023, 74(S1): 287-294.
[3]	张思雨, 殷勇高, 贾鹏琦, 叶威. 双U型地埋管群跨季节蓄热特性研究[J]. 化工学报, 2023, 74(S1): 295-301.
[4]	金正浩, 封立杰, 李舒宏. 氨水溶液交叉型再吸收式热泵的能量及分析[J]. 化工学报, 2023, 74(S1): 53-63.
[5]	李锦潼, 邱顺, 孙文寿. 煤浆法烟气脱硫中草酸和紫外线强化煤砷浸出过程[J]. 化工学报, 2023, 74(8): 3522-3532.
[6]	李艳辉, 丁邵明, 白周央, 张一楠, 于智红, 邢利梅, 高鹏飞, 王永贞. 非常规服役超临界锅炉的微纳尺度腐蚀动力学模型建立及应用[J]. 化工学报, 2023, 74(6): 2436-2446.
[7]	王承泽, 顾凯丽, 张晋华, 石建轩, 刘艺娓, 李锦祥. 硫化协同老化零价铁增效去除水中Cr（Ⅵ）的作用机制[J]. 化工学报, 2023, 74(5): 2197-2206.
[8]	郑杰元, 张先伟, 万金涛, 范宏. 丁香酚环氧有机硅树脂的制备及其固化动力学研究[J]. 化工学报, 2023, 74(2): 924-932.
[9]	周桓, 张梦丽, 郝晴, 吴思, 李杰, 徐存兵. 硫酸镁型光卤石转化钾盐镁矾的过程机制与动态规律[J]. 化工学报, 2022, 73(9): 3841-3850.
[10]	马语峻, 刘向军. 多孔陶瓷膜烟气水分回收理论与模型研究[J]. 化工学报, 2022, 73(9): 4103-4112.
[11]	戚子豪, 钟文琪, 陈曦, 周冠文, 赵小亮, 辛美静, 陈翼, 朱永长. 基于混合建模的水泥生料分解过程动态特性研究[J]. 化工学报, 2022, 73(5): 2039-2051.
[12]	张成, 潘立志, 李元. 基于加权统计特征KICA的故障检测与诊断方法[J]. 化工学报, 2022, 73(2): 827-837.
[13]	宋谦石, 王潇伟, 张威, 汪小憨, 李浩文, 乔瑜. 无机元素对生物质焦炭-CO₂气化反应性的催化/抑制作用研究及模型构建[J]. 化工学报, 2022, 73(11): 5240-5250.
[14]	王佳星, 赵家睿, 李斯特, 付媛, 刘凤杰, 韦昌. 不同放汽阀流量特性对弹射过程的影响规律[J]. 化工学报, 2021, 72(S1): 318-325.
[15]	张牧星, 张小松, 丁烨, 宋翼. 氧化石墨烯膜间距对电渗析空气除湿特性影响的分子动力学研究[J]. 化工学报, 2021, 72(S1): 63-69.