基于工况识别的污水处理过程多目标优化控制

doi:10.11949/0438-1157.20190453

摘要/Abstract

摘要：

针对污水处理过程中能耗大和罚款高等问题，设计了一种基于工况识别的污水处理智能优化控制系统。为保证工况识别的准确性和实时性，利用自适应遗传算法从多种入水参数中选取参考变量，然后基于建立的历史知识库，对入水实时工况进行识别。针对能耗和罚款的多目标优化问题，基于历史知识的引导，通过智能决策的方法从 $p a r e t o$ 解集中选出最优偏好解，并对知识库进行更新。利用国际基准仿真平台BSM1进行验证，结果表明所提方法有效利用了历史工况的最优解信息，提高了算法的收敛性，降低了计算成本，同时可将能耗和罚款控制在较低的范围。

关键词: 污水处理, 工况识别, 过程控制, 优化, 历史知识, 动态仿真

Abstract:

Aiming at the problems in wastewater treatment process, such as high energy consumption and penalty, a condition recognition based intelligent optimal control system for wastewater treatment is proposed. In order to ensure the accuracy and real-time performance of condition identification, the adaptive genetic algorithm is used to select reference variables from a variety of influent parameters, then based on the established historical knowledge base, identifies the real-time influent condition. Multi-objective optimization for energy consumption and penalty is guided by historical knowledge, and through the method of intelligent decision-making, the optimal preference solution is selected from pareto solution set, then update the knowledge base. The international benchmark simulation platform BSM1 is used to verify the results. The results show that the proposed method effectively utilizes the optimal solution information of historical conditions, improves the convergence of the algorithm, reduces the computational cost, and can control the energy consumption and fines at a lower level.

Key words: wastewater treatment, condition recognition, process control, optimization, historical knowledge, dynamic simulation

中图分类号:

TP 273

李永明, 史旭东, 熊伟丽. 基于工况识别的污水处理过程多目标优化控制[J]. 化工学报, 2019, 70(11): 4325-4336.

Yongming LI, Xudong SHI, Weili XIONG. Condition recognition based intelligent multi-objective optimal control for wastewater treatment[J]. CIESC Journal, 2019, 70(11): 4325-4336.

图/表 21

图1 总体控制结构图

Fig.1 General control structure

表1 控制器的控制结构

Table 1 Structure of controllers

项目	PID1	PID2	PID3
控制对象	$S O 4$	$S O 5$	$S N O 2$
操纵变量	$K L a 4$	$K L a 5$	$S N O 2$
设定值	$S O 4 s e t$	$S O 5 s e t$	$S N O 2 s e t$

表1 控制器的控制结构

Table 1 Structure of controllers

项目	PID1	PID2	PID3
控制对象	$S O 4$	$S O 5$	$S N O 2$
操纵变量	$K L a 4$	$K L a 5$	$S N O 2$
设定值	$S O 4 s e t$	$S O 5 s e t$	$S N O 2 s e t$

图2 选取参考变量流程

Fig.2 Process of selecting reference variables

图3 更新知识库流程

Fig.3 Process of updating knowledge base

图4 控制器设定值与能耗/罚款之间的软测量模型

Fig.4 Soft sensor model between controller settings and energy/penalty

图5 多目标优化流程

Fig.5 Process of multi-objective optimization

表2 AGA和MOPSO参数设置

Table 2 Parameter settings of AGA and MOPSO

AGA	种群大小		最大迭代次数	个体维数	自适应交叉变异系数
AGA	120		100	16	k ₁=0.5,k ₂=0.7,k ₃=0.02,k ₄=0.03
MOPSO	种群大小	最大迭代次数	粒子维数	粒子位置限制	速度限制	$p a r e t o$ 解集容量上限
MOPSO	150	100	3	同控制器设定值范围	0~0.3	100

表2 AGA和MOPSO参数设置

Table 2 Parameter settings of AGA and MOPSO

AGA	种群大小		最大迭代次数	个体维数	自适应交叉变异系数
AGA	120		100	16	k ₁=0.5,k ₂=0.7,k ₃=0.02,k ₄=0.03
MOPSO	种群大小	最大迭代次数	粒子维数	粒子位置限制	速度限制	$p a r e t o$ 解集容量上限
MOPSO	150	100	3	同控制器设定值范围	0~0.3	100

图6 三种天气下入水参考变量变化

Fig.6 Changes of influent reference variables in three different weather

表3 BPNN和LSSVM参数设置

Table 3 Parameter settings of BPNN and LSSVM

建模	BPNN			SVM	LSSVM
建模	网络结构	神经元个数	激活函数	正则化常数	核函数
参考变量与能耗/罚款建模	输入层、隐含层1、隐含层2、输出层	隐含层1个数20，隐含层2个数5	Sigmoid	1000	高斯核，带宽为2
控制器设定值建模与能耗/罚款建模	输入层、隐含层1、隐含层2、输出层	隐含层1个数10，隐含层2个数5	Sigmoid	11000	高斯核，带宽为1

图7 参考变量与能耗/罚款之间的建模效果对比

Fig.7 Comparisons of modeling effects between reference variables and energy/penalty

表4 参考变量与能耗/罚款之间的模型精度对比

Table 4 Comparisons of modeling accuracy between reference variables and energy/penalty

Model	OCI		EQI
Model	RMSE	RRMSE	RMSE	RRMSE
PLS	98.4453	0.0252	869.3646	0.1425
BPNN	138.5836	0.0354	784.6886	0.1286
SVM	152.0103	0.0389	593.0627	0.0972
GPR	89.9344	0.0230	683.8666	0.1121
LSSVM	71.8087	0.0184	433.9814	0.0711

图8 三种天气下知识库储存工况个数变化

Fig.8 Number changes of knowledge base storage conditions in three different weather

图9 控制器设定值与能耗/罚款之间的建模效果对比

Fig.9 Comparisons of modeling effects between controller settings and energy/penalty

表5 设定值与能耗/罚款之间的模型精度对比

Table 5 Comparisons of modeling accuracy between set value and energy/penalty

Model	OCI		EQI
Model	RMSE	RRMSE	RMSE	RRMSE
PLS	42.7990	1.35×10^-2	1.9910	3.81×10^-4
BPNN	86.8150	2.73×10^-2	1.9451	3.72×10^-4
SVM	12.5423	4.00×10^-3	1.4167	2.71×10^-4
GPR	1.2121	3.82×10^-4	0.1444	0.28×10^-4
LSSVM	1.1964	3.77×10 ^- ⁴	0.0843	0.16×10 ^- ⁴

图10 pareto解集中粒子位置

Fig.10 Positions of particle in pareto solution set

图13 多目标优化寻优过程示意图

Fig.13 Diagram of multi-objective optimization process

图11 成本函数随聚类个数的变化

Fig.11 Changes of cost function with number of clusters

图12 pareto前沿示意图

Fig.12 Diagram of pareto frontier

图14 晴朗天气各控制器的控制效果和误差示意图

Fig.14 Control effects and errors of each controller in dry weather

表6 三种天气下控制器的性能指标

Table 6 Performances indicators of each controller in three weather conditions

指标	晴天		雨天		暴雨天
指标	AAE	TP	AAE	TP	AAE	TP
PID1( $S O 4$ )	5.0×10^-2	0.8831	5.0×10^-2	0.8940	5.1×10^-2	0.8876
PID2( $S O 5$ )	7.5×10^-3	0.9880	7.6×10^-3	0.9887	7.3×10^-3	0.9886
PID3( $S N O 2$ )	5.9×10^-3	0.9932	5.8×10^-3	0.9928	5.9×10^-3	0.9930

表6 三种天气下控制器的性能指标

Table 6 Performances indicators of each controller in three weather conditions

指标	晴天		雨天		暴雨天
指标	AAE	TP	AAE	TP	AAE	TP
PID1( $S O 4$ )	5.0×10^-2	0.8831	5.0×10^-2	0.8940	5.1×10^-2	0.8876
PID2( $S O 5$ )	7.5×10^-3	0.9880	7.6×10^-3	0.9887	7.3×10^-3	0.9886
PID3( $S N O 2$ )	5.9×10^-3	0.9932	5.8×10^-3	0.9928	5.9×10^-3	0.9930

表7 三种天气下不同控制方法的能耗和罚款

Table 7 Energy and penalty under different control methods in three weather conditions

方法	晴天				雨天				暴雨天
方法	OCI			EQI	OCI		EQI		OCI		EQI
Openloop	3729.3	—	6590.3	—	3729.3	—	7701.6	—	3729.3	—	7244.0	—
CCS	3907.7	↑4.77%	6101.2	↓7.42%	3918.2	↑5.07%	7118.5	↓7.57%	3931.2	↑5.41%	6641.7	↓8.31%
SOOC	3819.7	↑2.37%	6205.8	↓5.83%	3830.9	↑2.72%	7215.2	↓6.32%	3830.9	↑2.72%	7215.2	↓0.40%
NSGAII	3705.6	↓0.64%	6466.8	↓1.87%	3723.8	↓0.14%	7369.0	↓4.32%	3694.3	↓0.94%	7055.3	↓2.60%
MOPSO	3656.3	↓1.95%	6348.7	↓3.67%	3676.3	↓1.42%	7461.3	↓3.12%	3673.6	↓1.49%	7089.8	↓2.13%
CRBMO	3303.3	↓11.42%	6273.8	↓4.80%	3275.4	↓12.17%	7407.1	↓3.82%	3313.6	↓11.15%	6878.9	↓5.04%

参考文献 32

1	Zhao J , Gui L , Wang Q , et al . Aged refuse enhances anaerobic digestion of waste activated sludge.[J]. Water Research, 2017, 123: 724-733.
2	Santín I , Pedret C , Vilanova R , et al . Advanced decision control system for effluent violations removal in wastewater treatment plants[J]. Control Engineering Practice, 2016, 49: 60-75.
3	Wang L , Qiang Z , Li Y , et al . An insight into the removal of fluoroquinolones in activated sludge process: sorption and biodegradation characteristics[J]. Journal of Environmental Sciences, 2017, 56(6): 263-271.
4	O’Brien M , Mack J , Lennox B , et al . Model predictive control of an activated sludge process: a case study[J]. Control Engineering Practice, 2015, 19(1): 54-61.
5	中华人民共和国生态环境部 . 2017 中国生态环境状况公报[EB/OL]. (2018-05-31)[2019-04-29].
	Ministry of Ecological Environment of the People's Republic of China. Bulletin on the State of Ecological Environment of China, 2017[EB/OL]. (2018-05-31)[2019-04-29].
6	Hreiz R , Latifi M A , Roche N . Optimal design and operation of activated sludge processes: state-of-the-art[J]. Chemical Engineering Journal, 2015, 281: 900-920.
7	魏伟, 王小艺, 王藩, 等 . 基于线性自抗扰的城市污水脱氮控制仿真[J]. 化工学报, 2016, 67(3): 1032-1039.
	Wei W , Wang X Y , Wang P , et al . Nitrogen removal in wastewater treatment processes based on linear active disturbance rejection control and its dynamic simulation[J]. CIESC Journal, 2016, 67(3): 1032-1039.
8	Oturan M A , Aaron J J . Advanced oxidation processes in water/wastewater treatment: principles and applications. A review[J]. Critical Reviews in Environmental Science & Technology, 2014, 44(23): 2577-2641.
9	Machado V C , Gabriel D , Lafuente J , et al . Cost and effluent quality controllers design based on the relative gain array for a nutrient removal WWTP[J]. Water Research, 2009, 43(20): 5129-5141.
10	Ostace G S , Baeza J A , Guerrero J , et al . Development and economic assessment of different WWTP control strategies for optimal simultaneous removal of carbon, nitrogen and phosphorus[J]. Computers & Chemical Engineering, 2013, 53: 164-177.
11	Zhang R , Xie W M , Yu H Q , et al . Optimizing municipal wastewater treatment plants using an improved multi-objective optimization method[J]. Bioresource Technology, 2014, 157: 161-165.
12	Piotrowski R , Brdys M A , Konarczak K , et al . Hierarchical dissolved oxygen control for activated sludge processes[J]. Control Engineering Practice, 2008, 16(1): 114-131.
13	Ostace G S , Baeza J A , Guerrero J , et al . Development and economic assessment of different WWTP control strategies for optimal simultaneous removal of carbon, nitrogen and phosphorus[J]. Computers & Chemical Engineering, 2013, 53(4): 164-177.
14	Machado V C , Gabriel D , Lafuente J , et al . Cost and effluent quality controllers design based on the relative gain array for a nutrient removal WWTP[J]. Water Research, 2009, 43(20): 5129-5141.
15	Dai H , Chen W , Lu X . The application of multi-objective optimization method for activated sludge process: a review[J]. Water Science & Technology, 2016, 73(2): 223-235.
16	Sweetapple C , Fu G , Butler D . Multi-objective optimisation of wastewater treatment plant control to reduce greenhouse gas emissions[J]. Water Research, 2014, 55: 52-62.
17	Chen W , Lu X , Yao C . Optimal strategies evaluated by multi-objective optimization method for improving the performance of a novel cycle operating activated sludge process[J]. Chemical Engineering Journal, 2015, 260: 492-502.
18	Hreiz R , Roche N , Benyahia B , et al . Multi-objective optimal control of small-size wastewater treatment plants[J]. Chemical Engineering Research & Design, 2015, 102: 345-353.
19	Santín I , Pedret C , Vilanova R . Applying variable dissolved oxygen set point in a two level hierarchical control structure to a wastewater treatment process[J]. Journal of Process Control, 2015, 28: 40-55.
20	周红标, 乔俊飞 . 混合多目标骨干粒子群优化算法在污水处理过程优化控制中的应用[J]. 化工学报, 2017, 68(9): 3511-3521.
	Zhou H B , Qiao J F . Optimal control of wastewater treatment process using hybrid multi-objective barebones particle swarm optimization algorithm[J]. CIESC Journal, 2017, 68(9): 3511-3521.
21	乔俊飞, 韩改堂, 周红标 . 基于知识的污水生化处理过程智能优化方法[J]. 自动化学报, 2017, 43(6): 1038-1046.
	Qiao J F , Hang G T , Zhou H B . Knowledge-based intelligent optimal control for wastewater biochemical treatment process[J]. Acta Automatica Sinica, 2017, 43(6): 1038-1046.
22	Francisco M , Vega P , Revollar S . Model predictive control of BSM1 benchmark of wastewater treatment process: a tuning procedure[C]//2011 50th IEEE Conference on Decision and Control and European Control Conference. IEEE, 2011: 7057-7062.
23	Santín I , Barbu M , Pedret C , et al . Control strategies for nitrous oxide emissions reduction on wastewater treatment plants operation[J]. Water Research, 2017, 125: 466-477.
24	Jeppsson U , Pons M N . The COST benchmark simulation model-current state and future perspective[J]. Control Engineering Practice, 2004, 12(3): 299-304.
25	颜晓娟, 龚仁喜, 张千锋 . 优化遗传算法寻优的SVM在短期风速预测中的应用[J]. 电力系统保护与控制, 2016, 44(9): 38-42.
	Yan X J , Gong R X , Zhang Q F . Application of optimization SVM based on improved genetic algorithm in short-term wind speed prediction[J]. Power System Protection and Control, 2016, 44(9): 38-42.
26	Li K , Kwong S , Zhang Q , et al . Interrelationship-based selection for decomposition multiobjective optimization[J]. IEEE Transactions on Cybernetics, 2015, 45(10): 2076-2088.
27	Morris G M , Goodsell D S , Halliday R S , et al . Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function[J]. Journal of Computational Chemistry, 2015, 19(14): 1639-1662.
28	Li M , Yang S , Liu X . Diversity comparison of Pareto front approximations in many-objective optimization[J]. IEEE Trans Cybern, 2017, 44(12): 2568-2584.
29	Wang G G , Gandomi A H , Alavi A H , et al . A hybrid method based on krill herd and quantum-behaved particle swarm optimization[J]. Neural Computing & Applications, 2016, 27(4): 989-1006.
30	Mouazen A M , Kuang B , de Baerdemaeker J , et al . Comparison among principal component, partial least squares and back propagation neural network analyses for accuracy of measurement of selected soil properties with visible and near infrared spectroscopy[J]. Geoderma, 2010, 158(1/2): 23-31.
31	Koriyama T , Nose T , Kobayashi T . Statistical parametric speech synthesis based on Gaussian process regression[J]. IEEE Journal of Selected Topics in Signal Processing, 2017, 8(2): 173-183.
32	Qiao J , Zhang W . Dynamic multi-objective optimization control for wastewater treatment process[J]. Neural Computing and Applications, 2018, 29(11): 1261-1271.

[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]	康飞, 吕伟光, 巨锋, 孙峙. 废锂离子电池放电路径与评价研究[J]. 化工学报, 2023, 74(9): 3903-3911.
[5]	何松, 刘乔迈, 谢广烁, 王斯民, 肖娟. 高浓度水煤浆管道气膜减阻两相流模拟及代理辅助优化[J]. 化工学报, 2023, 74(9): 3766-3774.
[6]	陈哲文, 魏俊杰, 张玉明. 超临界水煤气化耦合SOFC发电系统集成及其能量转化机制[J]. 化工学报, 2023, 74(9): 3888-3902.
[7]	曹跃, 余冲, 李智, 杨明磊. 工业数据驱动的加氢裂化装置多工况切换过渡状态检测[J]. 化工学报, 2023, 74(9): 3841-3854.
[8]	齐聪, 丁子, 余杰, 汤茂清, 梁林. 基于选择吸收纳米薄膜的太阳能温差发电特性研究[J]. 化工学报, 2023, 74(9): 3921-3930.
[9]	邢雷, 苗春雨, 蒋明虎, 赵立新, 李新亚. 井下微型气液旋流分离器优化设计与性能分析[J]. 化工学报, 2023, 74(8): 3394-3406.
[10]	陈国泽, 卫东, 郭倩, 向志平. 负载跟踪状态下的铝空气电池堆最优功率点优化方法[J]. 化工学报, 2023, 74(8): 3533-3542.
[11]	刘文竹, 云和明, 王宝雪, 胡明哲, 仲崇龙. 基于场协同和耗散的微通道拓扑优化研究[J]. 化工学报, 2023, 74(8): 3329-3341.
[12]	张曼铮, 肖猛, 闫沛伟, 苗政, 徐进良, 纪献兵. 危废焚烧处理耦合有机朗肯循环系统工质筛选与热力学优化[J]. 化工学报, 2023, 74(8): 3502-3512.
[13]	诸程瑛, 王振雷. 基于改进深度强化学习的乙烯裂解炉操作优化[J]. 化工学报, 2023, 74(8): 3429-3437.
[14]	吴文涛, 褚良永, 张玲洁, 谭伟民, 沈丽明, 暴宁钟. 腰果酚生物基自愈合微胶囊的高效制备工艺研究[J]. 化工学报, 2023, 74(7): 3103-3115.
[15]	汤晓玲, 王嘉瑞, 朱玄烨, 郑仁朝. 基于Pickering乳液的卤醇脱卤酶催化合成手性环氧氯丙烷[J]. 化工学报, 2023, 74(7): 2926-2934.