基于卷积层-注意力机制的长短期记忆网络出水氨氮浓度预测方法

doi:10.11949/0438-1157.20240599

摘要/Abstract

摘要：

为解决城市污水处理过程出水氨氮浓度难以实时精准测量的问题，本文构建了一种融合卷积层和注意力机制的长短期记忆网络（Convolutional Layer and Squeeze-and-Excitation Attention Mechanism based Long Short-Term Memory Network， CSA-LSTM）模型：首先，通过引入卷积层（Convolutional Layer， CL），深度提取数据中的非线性特征，并通过注意力机制（Squeeze-and-Excitation Attention Mechanism， SEAM）自适应分配特征通道的权重，实现特征解耦；其次，长短期记忆网络（Long Short-Term Memory network， LSTM）提取时间序列数据长期依赖关系，实现出水氨氮浓度的实时预测；然后，提出一种具有自适应采集函数的贝叶斯优化算法，实现模型参数优化，进一步提高模型精度；最后，基于基准实验和实际污水处理厂数据测试CSA-LSTM的有效性，结果表明模型具有较高的出水氨氮浓度预测精度，能够有效应对城市污水处理中数据的强非线性、耦合性以及时间依赖性问题，具有良好的泛化能力。

关键词: 城市污水处理, 氨氮浓度预测, 神经网络, 特征提取, 优化, 算法

Abstract:

To address the issue of accurately measuring the effluent ammonia nitrogen concentration in real-time during urban wastewater treatment processes, this paper constructs a Convolutional Layer (CL)and Squeeze-and-Excitation Attention Mechanism (SEAM)based Long Short-Term Memory Network (CSA-LSTM) model. Firstly, by introducing the CL, nonlinear features within the data are deeply extracted. The SEAM adaptively allocates weights to feature channels, achieving feature decoupling. Secondly, the Long Short-Term Memory network (LSTM) extracts long-term dependencies in time-series data to realize real-time prediction of effluent ammonia nitrogen concentration. Then, a Bayesian optimization algorithm with an adaptive acquisition function is proposed to optimize model parameters, further enhancing model accuracy. Finally, the effectiveness of CSA-LSTM is tested based on benchmark experiments and actual wastewater treatment plant (WWTP) data. The results show that the model has high predictive accuracy for effluent ammonia nitrogen concentration, can effectively handle the strong nonlinearity, coupling, and time-dependency problems in urban wastewater treatment data, and has good generalization ability

Key words: wastewater treatment plant, ammonia nitrogen concentration prediction, neural network, feature extraction, optimization, algorithm

中图分类号:

TP183

张勇, 赵景波, 权利敏. 基于卷积层-注意力机制的长短期记忆网络出水氨氮浓度预测方法[J]. 化工学报, DOI: 10.11949/0438-1157.20240599.

Yong ZHANG, Jingbo ZHAO, Limin QUAN. A prediction method for effluent ammonia nitrogen concentration based on convolutional layer and attention mechanism long short-term memory network[J]. CIESC Journal, DOI: 10.11949/0438-1157.20240599.

图/表 10

图1 LSTM最小单元结构图

Fig.1 Diagram of the minimum unit structure of LSTM

图2 相关变量自相关系数图

Fig.2 Autocorrelation coefficient plot for relevant variables

图3 CSA-LSTM模型架构

Fig.3 CSA-LSTM model architecture

表1 不同实验参数设置

Table 1 Parameter settings for different experiments

实验出水氨氮浓度预测	Mackey–Glass时间序列预测
SOFNN-HLA^[28]k_max=30, t_max=100, E _d =0.01, η= 0.001	k_max=30, t_max=100, E _d =0.01, η= 0.001
PDF-FNN^[27t_max = 10⁴, η = 8 × 10^-4, µ = 0.001, ε = 0.001, σ_g = 0.05, n = 4	t_max = 10⁴, η = 8 × 10^-4, µ = 0.001, ε = 0.001, σ_g = 0.05, r = 4
SEAM-LSTM t_max=900, η= 0.001, r = 4, n=6, m=2	t_max=700, η= 0.001, r = 4, n=6, m=2
CL-LSTM t_max=850, η= 0.001, r = 4, n=6, m=2	t_max=600, η= 0.001, r = 4, n=6, m=2
CSA-LSTM t_max=1000, η= 0.001, r = 4, n=6, m=2	t_max=800, η= 0.001, r = 4, n=6, m=2

图4 Mackey–Glass预测R2图

Fig.4 R2 graph of Mackey-Glass prediction

表2 Mackey–Glass时间序列预测对比结果

Table 2 Comparative results of Mackey-Glass time series prediction

方法	NRMSE	R²	APE
SOFNN-HLA^[26]	0.0126	0.99013	0.494
PDF-FNN^[25]	0.0265	0.98651	0.517
SEAM-LSTM	0.0324	0.97012	0.891
CL-LSTM	0.0262	0.98152	0.775
本文方法	0.0091	0.99714	0.483

图5 出水氨氮浓度预测模型结构框图

Fig.5 Structure of the effluent ammonia nitrogen concentration prediction model

图6 不同模型预测结果对比图

Fig.6 Comparison chart of prediction results for various models

图7 不同模型预测误差图

Fig.7 Prediction error graph of different models

表3 出水氨氮浓度预测对比结果

Table 3 Comparison of predicted and actual ammonia nitrogen concentration in effluent

方法	NRMSE	R²	APE
SOFNN-HLA^[26]	0.0432	0.98981	0.771
PDF-FNN^[25]	0.0453	0.98723	0.837
SEAM-LSTM	0.1292	0.96941	1.524
CL-LSTM	0.1012	0.97352	1.314
本文方法	0.0210	0.99334	0.592

参考文献 33

1	韩红桂, 赵雅倩, 杨宏燕, 等. 数据驱动的污水处理曝气过程低碳优化控制方法[J]. 北京工业大学学报, 2024, 50(2): 131-139.
	Han H G, Zhao Y Q, Yang H Y, et al. Data-driven optimal control method of low-carbon for waste water treatment aeration process[J]. Journal of Beijing University of Technology, 2024, 50(2): 131-139.
2	Wang J H, Guo Y, Peng S, et al. Prediction of effluent ammonia nitrogen in wastewater treatment plant based on self-organizing hybrid neural network[J]. Journal of Water Process Engineering, 2024, 59: 104930.
3	刘壮壮, 鞠然, 刘崇涛, 等. 电化学膜生物反应器处理污水性能提升策略及研究现状[J]. 化工学报, 2023, 74(11): 4433-4444.
	Liu Z Z, Ju R, Liu C T, et al. Strategies for performance enhancement of electrochemical membrane bioreactors for wastewater treatment and current research status[J]. CIESC Journal, 2023, 74(11): 4433-4444.
4	Mardani S, Baghdadi M, Torabian A, et al. Optimization of ammonia and COD removal from municipal wastewater effluent by electrochemical continuous flow reactor equipped with Ti/RuO2 and Cu foam[J]. Journal of Water Process Engineering, 2023, 55: 104185.
5	Zhang X Y, Tian J L, Jiang Y S, et al. Direct ammonium recovery from the permeate of a pilot-scale anaerobic MBR by biochar to advance low-carbon municipal wastewater reclamation and urban agriculture[J]. Science of the Total Environment, 2023, 877: 162872.
6	Wang G P, Dai X H, Zhao S H, et al. Research on ammonia removal from reject water produced from anaerobic digestion of thermally hydrolyzed sludge through partial nitrification—anammox[J]. Water, Air, & Soil Pollution, 2022, 233(4): 106.
7	蒙西, 乔俊飞, 韩红桂. 基于类脑模块化神经网络的污水处理过程关键出水参数软测量[J]. 自动化学报, 2019, 45(5): 906-919.
	Meng X, Qiao J F, Han H G. Soft measurement of key effluent parameters in wastewater treatment process using brain-like modular neural networks[J]. Acta Automatica Sinica, 2019, 45(5): 906-919.
8	Quan L M, Meng X, Qiao J F. Robust self-constructing fuzzy neural network-based online estimation for industrial product quality[J]. IEEE Transactions on Industrial Informatics, 2024, 20(2): 2213-2222.
9	Guo X, Li W J, Qiao J F. A self-organizing modular neural network based on empirical mode decomposition with sliding window for time series prediction[J]. Applied Soft Computing, 2023, 145: 110559.
10	Shin C, Szczuka A, Jiang R J, et al. Optimization of reverse osmosis operational conditions to maximize ammonia removal from the effluent of an anaerobic membrane bioreactor[J]. Environmental Science: Water Research & Technology, 2021, 7(4): 739-747.
11	Shahar B, Guttman L. An integrated, two-step biofiltration system with Ulva fasciata for sequenced removal of ammonia and nitrate in mariculture effluents[J]. Algal Research, 2020, 52: 102120.
12	张璐, 张嘉成, 韩红桂, 等. 基于模糊神经网络的污水处理生化除磷过程控制[J]. 化工学报, 2020, 71(3): 1217-1225.
	Zhang L, Zhang J C, Han H G, et al. FNN-based process control for biochemical phosphorus in WWTP[J]. CIESC Journal, 2020, 71(3): 1217-1225.
13	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.
	Li Y M, Shi X D, Xiong W L. Condition recognition based intelligent multi-objective optimal control for wastewater treatment[J]. CIESC Journal, 2019, 70(11): 4325-4336.
14	Zhou H B, Qiao J F. Soft sensing of effluent ammonia nitrogen using rule automatic formation-based adaptive fuzzy neural network[J]. Desalination and Water Treatment, 2019, 140: 132-142.
15	Qiao J F, Zhang Z Z, Bo Y C. An online self-adaptive modular neural network for time-varying systems[J]. Neurocomputing, 2014, 125: 7-16.
16	韩红桂, 陈治远, 乔俊飞, 等. 基于区间二型模糊神经网络的出水氨氮软测量[J]. 化工学报, 2017, 68(3): 1032-1040.
	Han H G, Chen Z Y, Qiao J F, et al. Soft-sensor method for effluent ammonia nitrogen based on interval type-2 fuzzy neural networks[J]. CIESC Journal, 2017, 68(3): 1032-1040.
17	Chantilas A, El-Khattabi A R, Gvino E, et al. Interlinkages and gaps: a review of the literature on intergovernmental relations for flood management in the face of climate change[J]. Frontiers in Sustainable Cities, 2023, 5: 1135513.
18	Dube P J, Vanotti M B, Szogi A A, et al. Enhancing recovery of ammonia from swine manure anaerobic digester effluent using gas-permeable membrane technology[J]. Waste Management, 2016, 49: 372-377.
19	Tahmasbi R, Kholghi M, Najarchi M, et al. Post-treatment of reclaimed municipal wastewater through unsaturated and saturated porous media in a large-scale experimental model[J]. Water, 2022, 14(7): 1137.
20	Moradvandi A, Abraham E, Goudjil A, et al. An identification algorithm of switched Box-Jenkins systems in the presence of bounded disturbances: an approach for approximating complex biological wastewater treatment models[J]. Journal of Water Process Engineering, 2024, 60: 105202.
21	Qin J, Guo H J. Expert system development on on-line measurement of sewage treatment based process[J]. Sensors \& Transducers, 2014, 164: 227-232.
22	Mirikitani D T, Nikolaev N. Recursive Bayesian recurrent neural networks for time-series modeling[J]. IEEE Transactions on Neural Networks, 2010, 21(2): 262-274.
23	Piri J, Pirzadeh B, Keshtegar B, et al. Reliability analysis of pumping station for sewage network using hybrid neural networks - genetic algorithm and method of moment[J]. Process Safety and Environmental Protection, 2021, 145: 39-51.
24	Waqas S, Harun N Y, Sambudi N S, et al. SVM and ANN modelling approach for the optimization of membrane permeability of a membrane rotating biological contactor for wastewater treatment[J]. Membranes, 2022, 12(9): 821.
25	Li D, Yang C H, Li Y G. A multi-subsystem collaborative Bi-LSTM-based adaptive soft sensor for global prediction of ammonia-nitrogen concentration in wastewater treatment processes[J]. Water Research, 2024, 254: 121347.
26	Farhi N, Kohen E, Mamane H, et al. Prediction of wastewater treatment quality using LSTM neural network[J]. Environmental Technology & Innovation, 2021, 23: 101632.
27	Qiao J F, Quan L M, Yang C L. Design of modeling error PDF based fuzzy neural network for effluent ammonia nitrogen prediction[J]. Applied Soft Computing, 2020, 91: 106239.
28	Meng X, Zhang Y, Quan L M, et al. A self-organizing fuzzy neural network with hybrid learning algorithm for nonlinear system modeling[J]. Information Sciences, 2023, 642: 119145.
29	Yaqub M, Asif H, Kim S, et al. Modeling of a full-scale sewage treatment plant to predict the nutrient removal efficiency using a long short-term memory (LSTM) neural network[J]. Journal of Water Process Engineering, 2020, 37: 101388.
30	Bhadra S, Sagan V, Skobalski J, et al. End-to-end 3D CNN for plot-scale soybean yield prediction using multitemporal UAV-based RGB images[J]. Precision Agriculture, 2024, 25(2): 834-864.
31	Yu H H, Yang L, Li D L, et al. A hybrid intelligent soft computing method for ammonia nitrogen prediction in aquaculture[J]. Information Processing in Agriculture, 2021, 8(1): 64-74.
32	Li J C, Lin S J, Zhang L, et al. Brain-inspired multimodal approach for effluent quality prediction using wastewater surface images and water quality data[J]. Frontiers of Environmental Science & Engineering, 2023, 18(3): 31.
33	Eledum H, Awadallah H H. A Monte Carlo study for dealing with multicollinearity and autocorrelation problems in linear regression using two stage ridge regression method[J]. Mathematics and Statistics, 2021, 9(5): 630-638.

[1]	李子扬, 郑楠, 方嘉宾, 魏进家. 再压缩S-CO₂布雷顿循环性能分析及多目标优化[J]. 化工学报, 2024, 75(6): 2143-2156.
[2]	张晗, 张淑宁, 刘珂, 邓冠龙. 基于慢特征分析与最小二乘支持向量回归集成的草酸钴合成过程粒度预报[J]. 化工学报, 2024, 75(6): 2313-2321.
[3]	许茹枫, 陈煜成, 高丹, 焦静雨, 高栋, 王海彬, 姚善泾, 林东强. 离子交换层析分离单抗电荷异质体的模型辅助过程优化[J]. 化工学报, 2024, 75(5): 1903-1911.
[4]	黄志鸿, 周利, 柴士阳, 吉旭. 耦合加氢装置优化的多周期氢网络集成[J]. 化工学报, 2024, 75(5): 1951-1965.
[5]	刘梦绮, 王凯, 骆广生. 基于人工智能的微分散基础研究[J]. 化工学报, 2024, 75(4): 1096-1104.
[6]	曾玉娇, 肖炘, 杨刚, 张意博, 郑光明, 李防, 汪凤玲. 基于机理与数据混合驱动的湿法磷酸生产过程代理建模与优化[J]. 化工学报, 2024, 75(3): 936-944.
[7]	陈思睿, 毕景良, 王雷, 李元媛, 陆规. 气液两相流流型特征无监督提取的卷积自编码器：机理及应用[J]. 化工学报, 2024, 75(3): 847-857.
[8]	张领先, 刘斌, 邓琳, 任宇航. 基于改进TSO优化Xception的PEMFC故障诊断[J]. 化工学报, 2024, 75(3): 945-955.
[9]	董益斌, 熊敬超, 王敬宇, 汪守康, 王亚飞, 黄群星. 融合激光雷达料位测算的锅炉燃烧优化模型预测控制[J]. 化工学报, 2024, 75(3): 924-935.
[10]	陈彦松, 阮达, 刘渊博, 郑通, 张帅帅, 马学虎. 微通道换热器拓扑结构优化与性能研究[J]. 化工学报, 2024, 75(3): 823-835.
[11]	蒙西, 王岩, 孙子健, 乔俊飞. 基于注意力模块化神经网络的城市固废焚烧过程氮氧化物排放预测[J]. 化工学报, 2024, 75(2): 593-603.
[12]	肖拥君, 时兆翀, 万仁, 宋璠, 彭昌军, 刘洪来. 反向传播神经网络用于预测离子液体的自扩散系数[J]. 化工学报, 2024, 75(2): 429-438.
[13]	李文俊, 赵中阳, 倪震, 周灿, 郑成航, 高翔. 基于气-液传质强化的湿法烟气脱硫CFD模拟研究[J]. 化工学报, 2024, 75(2): 505-519.
[14]	尹刚, 钱中友, 曹文琦, 全鹏程, 许亨权, 颜非亚, 王民, 向禹, 向冬梅, 卢剑, 左玉海, 何文, 卢润廷. 基于Adaboost-PSO-SVM的铝电解槽健康状态诊断方法研究[J]. 化工学报, 2024, 75(1): 354-365.
[15]	余洋, 罗祎青, 魏荣辉, 张文慧, 袁希钢. 考虑节点中断风险的弹性供应链设计方法[J]. 化工学报, 2024, 75(1): 338-353.