Adaptive prediction for effluent quality of wastewater treatment plant: Improvement with a dual-stage attention-based LSTM network

Tong An; Kuanliang Feng; Peijin Cheng; Ruojia Li; Zihao Zhao; Xiangyang Xu; Liang Zhu

doi:10.1016/j.jenvman.2024.120887

Adaptive prediction for effluent quality of wastewater treatment plant: Improvement with a dual-stage attention-based LSTM network

J Environ Manage. 2024 Apr 26:359:120887. doi: 10.1016/j.jenvman.2024.120887. Online ahead of print.

Authors

Tong An¹, Kuanliang Feng², Peijin Cheng¹, Ruojia Li¹, Zihao Zhao³, Xiangyang Xu⁴, Liang Zhu⁵

Affiliations

¹ College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, 310058, China.
² Zhejiang Supcon Information Technology Co., Ltd, Hangzhou, 310052, China.
³ Shanghai Municipal Engineering Design Institute (group) Co., Ltd, Shanghai, 200092, China.
⁴ College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, 310058, China; Zhejiang Provincial Engineering Laboratory of Water Pollution Control, Hangzhou, 310058, China.
⁵ College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, 310058, China; Innovation Center of Yangtze River Delta, Zhejiang University, Jiashan, 314100, China; Zhejiang Provincial Engineering Laboratory of Water Pollution Control, Hangzhou, 310058, China. Electronic address: felix79cn@zju.edu.cn.

PMID: 38678908
DOI: 10.1016/j.jenvman.2024.120887

Abstract

The accurate effluent prediction plays a crucial role in providing early warning for abnormal effluent and achieving the adjustment of feedforward control parameters during wastewater treatment. This study applied a dual-staged attention mechanism based on long short-term memory network (DA-LSTM) to improve the accuracy of effluent quality prediction. The results showed that input attention (IA) and temporal attention (TA) significantly enhanced the prediction performance of LSTM. Specially, IA could adaptively adjust feature weights to enhance the robustness against input noise, with R² increased by 13.18%. To promote its long-term memory ability, TA was used to increase the memory span from 96 h to 168 h. Compared to a single LSTM model, the DA-LSTM model showed an improvement in prediction accuracy by 5.10%, 2.11%, 14.47% for COD, TP, and TN. Additionally, DA-LSTM demonstrated excellent generalization performance in new scenarios, with the R² values for COD, TP, and TN increasing by 22.67%, 20.06%, and 17.14% respectively, while the MAPE values decreased by 56.46%, 63.08%, and 42.79%. In conclusion, the DA-LSTM model demonstrated excellent prediction performance and generalization ability due to its advantages of feature-adaptive weighting and long-term memory focusing. This has forward-looking significance for achieving efficient early warning of abnormal operating conditions and timely management of control parameters.

Keywords: Deep learning; Dual-stage attention-based LSTM; Effluent quality prediction; Multivariate time series; Wastewater treatment plants.