Modeling of wastewater reverse osmosis process based on Alopex evolutionary algorithm and ensemble learning multi-model back propagation neural network

doi:10.11949/0438-1157.20241085

Abstract

Abstract:

This paper proposes an AdaBoost strategy based on training quality and a multimodal neural network machine learning method based on Alopex evolutionary algorithm for the dynamic process of ultrafiltration performance in sewage reverse osmosis water pretreatment. Firstly, this paper constructs a class of generalized Bayesian inference probability indicators suitable for arbitrary distributions to classify multimodal states such as filtration and backwashing in ultrafiltration reverse osmosis wastewater treatment processes. Then, a neural network algorithm based on Alopex evolutionary algorithm and distribution-based AdaBoost ensemble strategy is used to model each modal process separately. Finally, the constructed generalized Bayesian inference-based probability indicators for each modality are used to integrate multiple models of the multimodal process. In order to verify the effectiveness of the method proposed in this paper, the method was applied to a two-year dataset collected from a community in the United States. The proposed method has good predictive performance for membrane resistance and backwash efficiency, and can effectively predict expected water quality changes.

Key words: ultrafiltration RO feed pre-treatment, backwash efficiency, hydraulic membrane resistance, back propagation neural network

摘要：

面向污水反渗透给水预处理中超滤性能动态过程，提出一种基于训练质量的AdaBoost策略和基于Alopex进化算法的多模态神经网络机器学习方法。首先，构建了一类适用于任意分布的广义贝叶斯推断概率指标对超滤反渗透污水处理过程中过滤和反洗等多模态状态进行分类，然后使用基于Alopex进化算法和基于分布的AdaBoost集成策略的神经网络算法针对每一个模态过程分别建模，最后利用构建的每个模态的基于广义贝叶斯推理的概率指标，将多模态的多个模型集成。为了验证所提方法的有效性，将该方法应用到美国某社区采集的两年数据集中，结果表明该方法对膜阻力和反洗效率具有很好的预测性能，能够很好地预测预期水质变化性能。

关键词: 超滤反渗透过程, 反冲洗效率, 水力膜阻力, 反向传播神经网络

CLC Number:

O 242

Yang ZHOU, Dan LIANG, Kai WANG, Yilan ZHANG, Li JIA. Modeling of wastewater reverse osmosis process based on Alopex evolutionary algorithm and ensemble learning multi-model back propagation neural network[J]. CIESC Journal, 2025, 76(4): 1661-1670.

周洋, 梁聃, 汪恺, 张艺蓝, 贾立. 基于Alopex进化算法和集成学习的多模态神经网络的污水反渗透过程建模[J]. 化工学报, 2025, 76(4): 1661-1670.

Figures/Tables 9

Fig.1 Simplified diagram of UF-RO seawater desalination system

Fig.2 Diagram of operation cycle (filtration/backwash)

Fig.3 Development flowchart of the proposed model

Fig.4 Flowchart of proposed method

Fig.5 Number of datasets for the first seven attributes

Table 1 Comparison of performance indicators for selecting attribute sets， backwashing efficiency， filtering， and resistance prediction after backwashing for 30 test datasets

方法	预测性能指标	常规操作条件				暴风雨期间
方法	预测性能指标	过滤膜阻力	反洗膜阻力	反洗效率	<R_PB>/<R_PB>₀	过滤膜阻力	反洗膜阻力	反洗效率	<R_PB>/<R_PB>₀
本文方法	R²	0.962	0.948	0.942	0.947	0.945	0.934	0.924	0.9251
本文方法	RMSE	0.027	0.031	0.013	0.013	0.039	0.053	0.012	0.016
本文方法-non	R²	0.911	0.896	0.891	0.898	0.893	0.882	0.853	0.858
本文方法-non	RMSE	0.037	0.041	0.023	0.026	0.047	0.066	0.024	0.026
随机森林	R²	0.859	0.849	0.839	0.844	0.846	0.831	0.814	0.818
随机森林	RMSE	0.067	0.071	0.053	0.052	0.078	0.091	0.053	0.059
LSTM	R²	0.881	0.866	0.863	0.869	0.864	0.852	0.840	0.841
LSTM	RMSE	0.052	0.056	0.038	0.037	0.063	0.077	0.038	0.042

Fig.6 Taking the first 10 cycles of the 142 cycles in the test dataset, the proposed integrated BPNN-AEA model compares the predicted and observed values (above the horizontal dashed line represents the filtering mode, and below the horizontal dashed line represents the backwashing process)

Fig.7 Taking the constant coagulant dosage (2.7 mg/L Fe3+) test dataset as an example, verify the prediction accuracy of the proposed model (the range of ultrafiltration feed turbidity and chlorophyll-a content is 0.6—4.4 NTU and 58—244 µg/L, respectively)

Fig.8 Taking the test dataset (512 cycles, 8 d of operation time) with stormy weather as an example, the prediction accuracy of the model's backwash efficiency is proposed (the range of ultrafiltration feed turbidity and chlorophyll-a content is 1.5—19 NTU and 43—142 µg/L, respectively)

References 30

1	Elimelech M, Phillip W A. The future of seawater desalination: energy, technology, and the environment[J]. Science, 2011, 333(6043): 712.
2	Villacorte L O, Tabatabai S A A, Anderson D M, et al. Seawater reverse osmosis desalination and (harmful) algal blooms[J]. Desalination, 2015, 360: 61.
3	Henthorne L, Boysen B. State-of-the-art of reverse osmosis desalination pretreatment[J]. Desalination, 2015, 356: 129.
4	Chua K T, Hawlader M N A, Malek A. Pretreatment of seawater: results of pilot trials in Singapore[J]. Desalination, 2003, 159(3): 225.
5	Schurer R, Janssen A, Villacorte L, et al. Performance of ultrafiltration and coagulation in an UF-RO seawater desalination demonstration plant[J]. Desalination and Water Treatment, 2012, 42(1/2/3): 57.
6	Brehant A, Bonnelye V, Perez M. Comparison of MF/UF pretreatment with conventional filtration prior to RO membranes for surface seawater desalination[J]. Desalination, 2002, 144(1/2/3): 353.
7	罗先国. 水环境工程污水处理技术研讨[J]. 地质研究与环境保护, 2024, 3(7): 83.
	Luo X G. Discussion on wastewater treatment technology in water environment engineering[J]. Geological Research and Environmental Protection, 2024, 3(7): 83.
8	Chang H, Liu B, Yang P, et al. Salt backwashing of organic-fouled ultrafiltration membranes: effects of feed water properties and hydrodynamic conditions[J]. Journal of Water Process Engineering, 2019, 30: 100429.
9	曹永红. 反渗透水处理设备在工业污水处理中的应用分析[J]. 工程建设, 2020, 3(9): 126.
	Cao Y H. Application analysis of reverse osmosis water treatment equipment in industrial wastewater treatment[J]. Engineering Construction, 2020, 3(9): 126.
10	刘勇超, 陈启斌, 王朝旭, 等. 外加碳源对生物炭基潜流人工湿地净化污水处理厂尾水效能的影响[J]. 环境工程技术学报, 2023, 13(4): 1295.
	Liu Y C, Chen Q B, Wang C X, et al. Effect of external carbon addition on pollutants removal from the tail water of a sewage treatment plant by biochar-based subsurface flow constructed wetland[J]. Journal of Environmental Engineering Technology, 2023, 13(4): 1295.
11	Woo Y C, Lee J J, Tijing L D, et al. Characteristics of membrane fouling by consecutive chemical cleaning in pressurized ultrafiltration as pre-treatment of seawater desalination[J]. Desalination, 2015, 369: 51.
12	He Y, Sharma J, Bogati R, et al. Impacts of aging and chemical cleaning on the properties and performance of ultrafiltration membranes in potable water treatment[J]. Separation Science and Technology, 2014, 49(9): 1317.
13	Tabatabai S A A, Kennedy M D, Amy G L, et al. Optimizing inline coagulation to reduce chemical consumption in MF/UF systems[J]. Desalination and Water Treatment, 2009, 6(1/2/3): 94.
14	Zhou Y, Li S. BP neural network modeling with sensitivity analysis on monotonicity based Spearman coefficient[J]. Chemometrics and Intelligent Laboratory Systems, 2020, 200: 103977.
15	Zhou Y, Li S, Xiong N. Improved vine copula-based dependence description for multivariate process monitoring based on ensemble learning[J]. Industrial & Engineering Chemistry Research, 2019, 58(9): 3782.
16	Zhou Y, Li S. A probabilistic Copula-based fault detection method with TrAdaBoost strategy for industrial IoT[J]. IEEE Internet of Things Journal, 2022, 10(9): 7813.
17	Zhou Y, Ren X, Li S. Enhancing quality of multivariate process monitoring based on vine copula and active learning strategy[J]. Industrial & Engineering Chemistry Research, 2018, 57(23): 7961.
18	Gille D, Czolkoss W. Ultrafiltration with multi-bore membranes as seawater pre-treatment[J]. Desalination, 2005, 182(1/2/3): 301.
19	闫霞. 电厂水处理存在问题及对策分析[J]. 工程管理与技术探讨, 2025, 7(1): 203.
	Yan X. Analysis of problems and countermeasures in water treatment of power plants[J]. Engineering Management and Technology Discussion, 2025, 7(1): 203.
20	Chew J W, Kilduff J, Belfort G. The behavior of suspensions and macromolecular solutions in crossflow microfiltration: an update[J]. Journal of Membrane Science, 2020, 601: 117865.
21	Carrão H, Naumann G, Barbosa P. Mapping global patterns of drought risk: an empirical framework based on sub-national estimates of hazard, exposure and vulnerability[J]. Global Environmental Change, 2016, 39: 108.
22	Wolf P H, Siverns S, Monti S. UF membranes for RO desalination pretreatment[J]. Desalination, 2005, 182(1/2/3): 293.
23	Diaz S D, Peña L V, Cabrera E G, et al. Effect of previous coagulation in direct ultrafiltration of primary settled municipal wastewater[J]. Desalination, 2012, 304: 41.
24	Lee J D, Lee S H, Jo M H, et al. Effect of coagulation conditions on membrane filtration characteristics in coagulation-microfiltration process for water treatment[J]. Environmental Science & Technology, 2000, 34(17): 3780.
25	Li Z, Yan H, Zhang C, et al. Long-short term spatiotemporal tensor prediction for passenger flow profile[J]. IEEE Robotics and Automation Letters, 2020, 5(4): 5010.
26	Zhang J, Agustriyanto R. Multivariable inferential feed-forward control[J]. Industrial & Engineering Chemistry Research, 2003, 42(18): 4186.
27	Zhou Y, Ren X, Li S. Probabilistic weighted copula regression model with adaptive sample selection strategy for complex industrial processes[J]. IEEE Transactions on Industrial Informatics, 2020, 16(11): 6972.
28	Li S, Li F. Alopex-based evolutionary algorithm and its application to reaction kinetic parameter estimation[J]. Computers & Industrial Engineering, 2011, 60(2): 341.
29	Li R, Shang Z, Zheng C, et al. Efficient distributional reinforcement learning with Kullback-Leibler divergence regularization[J]. Applied Intelligence, 2023, 53(21): 24847.
30	张云, 陈建标, 杨柳俊. 连续微滤+反渗透技术在印染废水回用中的研究[J]. 长江科学院院报, 2013, 30(11): 24.
	Zhang Y, Chen J B, Yang L J. Reuse of dyeing waste water by CMF and RO combined treatment[J]. Journal of Changjiang River Scientific Research Institute, 2013, 30(11): 24.