污水處理廠放流水中硝酸鹽氮（NO3-N）與銨氮（NH4-N）濃度受生物反應機制、操作條件、時間延遲及多源感測資訊等因素共同影響，具有高度非線性與時空相依特性，使傳統監測方式難以往前掌握未來水質變化，影響操作決策與放流水品質管理。為提升含氮濃度之提前預測能力，本研究提出一套結合時空圖注意力機制（Spatio-Temporal Graph Attention）與活性污泥模型第一版（Activated Sludge Model No. 1, ASM1）之混合預測方法，藉由融合資料驅動模型之時空特徵學習能力與機制模型之生物反應知識，以提升短期及多步長預測之準確性與模型穩健性。

本研究以荷蘭 Tilburg 污水處理廠連續 270 天之營運資料作為實驗資料，建立符合時間因果關係之資料處理與模型驗證流程，完成 Attention、GCN-Attention、ASM1 輔助殘差模型及 ASM1-GCN 混合模型等技術架構之實作，並與 GRU、LSTM、TCN、圖卷積網路（GCN）及校正後 ASM1 等模型進行系統性比較。為確保研究結果之可信度與泛化能力，本研究採用五折滑動時間序列交叉驗證、多組隨機種子（41、42、43）重複實驗、簡單基準模型比較、受控診斷實驗及運算資源代理評估等方式，全面檢驗各模型於不同預測步長下之預測效能、初始化穩定性及運算成本。

研究結果顯示，在 1 小時提前預測中，GRU 具有最佳整體預測表現，平均加權誤差（N_{wMAE}）為 22.74，且於不同隨機種子下皆維持最佳穩定性；在 6 小時提前預測中，ASM1-GCN-LSTM 混合模型之平均誤差最低（N_{wMAE}=50.32），惟與 TCN（N_{wMAE}=51.38）之差異有限。進一步分析發現，Attention 機制之效益會隨模型架構及預測步長而改變，未能穩定提升預測準確度；另一方面，拓撲證偽實驗亦未證實物理拓撲資訊具有一致性的精度優勢，顯示 GCN 與 ASM1 更適合作為提升模型可解釋性與可審計性的空間及機制知識來源，而非單純追求預測精度。

本研究建立一套整合時序資訊、空間拓撲、注意力機制與活性污泥機制模型之混合預測與驗證架構，不僅提供不同預測提前量、污染物類型及運算限制下之模型選擇依據，亦建立完整且可重現之模型驗證流程，可協助污水處理業者提前掌握放流水含氮濃度變化，提升水質預警能力、模型選用效率及智慧化操作決策品質。

Effluent nitrate (NO_3mathrm{-}N) and ammonium (NH_4mathrm{-}N) concentrations in wastewater treatment plants (WWTPs) exhibit strong nonlinearity and spatiotemporal dependencies, making lead-time water quality forecasting challenging. This study proposes a hybrid forecasting framework integrating Spatio-Temporal Graph Attention and Activated Sludge Model No. 1 (ASM1) to combine data-driven feature learning with biological process mechanisms. Using 270 consecutive days of operational data from the Tilburg WWTP, 18 models were systematically evaluated across 1-hour and 6-hour forecast horizons under five-fold walk-forward time-series cross-validation and multi-seed trials. Experimental results demonstrate that for 1-hour forecasting, GRU achieves the best overall performance (N_{wMAE}=22.74) and highest stability. For 6-hour forecasting, the ASM1-GCN-LSTM hybrid yields the lowest mean error (N_{wMAE}=50.32), closely followed by TCN (N_{wMAE}=51.38). Furthermore, attention mechanisms and physical graph topology do not yield consistent accuracy gains, indicating that GCN and ASM1 are most valuable as structured domain-knowledge sources that enhance interpretability and auditability rather than numerical accuracy alone. This study establishes a reproducible validation framework and practical model-selection guidelines to support proactive early warning and smart operational decision-making in WWTPs.

Allen, L. and Cordiner, J. Knowledge-enhanced data-driven modeling of wastewater treatment processes for energy consumption prediction. Computers & Chemical Engineering, 2025. https://doi.org/10.1016/j.compchemeng.2024.108982

An, T., Feng, K., Cheng, P., Li, R., Zhao, Z., Xu, X., and Zhu, L. Adaptive prediction for effluent quality of wastewater treatment plant: Improvement with a dual-stage attention-based LSTM network. Journal of Environmental Management, 359:120887, 2024. https://doi.org/10.1016/j.jenvman.2024.120887

Bai, S., Kolter, J. Z., and Koltun, V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv preprint arXiv:1803.01271, 2018. https://doi.org/10.48550/arXiv.1803.01271

Brun, R., Reichert, P., and Künsch, H. R. Practical identifiability analysis of large environmental simulation models. Water Resources Research, 37(4):1015–1030, 2001. https://doi.org/10.1029/2000WR900350

Busch, J., Elixmann, D., Kühl, P., Gerkens, C., Schlöder, J. P., Bock, H. G., and Marquardt, W. State estimation for large-scale wastewater treatment plants. Water Research, 47(13):4774–4787, 2013. https://doi.org/10.1016/j.watres.2013.04.007

Chen, H., Zhang, Q., Xie, J., Wang, K., and Yue, W. Multi-pollutant prediction and process parameter optimization of a wastewater treatment plant based on machine learning models. Frontiers in Water, 8:1719097, 2026. https://doi.org/10.3389/frwa.2026.1719097

Cheng, Z., Xia, W., Zhu, J.-J., Cao, J., Ren, Z. J., and Yuan, H. A comprehensive guideline for hybrid modeling of engineered microbial processes. Water Research, 2025. https://doi.org/10.1016/j.watres.2025.124559

Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., and Bengio, Y. (2014). Learning phrase representations using RNN encoder-decoder for statistical machine translation. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1724–1734. https://doi.org/10.3115/v1/d14-1179

Cilibiu, C., Albu, H. C., and Abrudan, A.-C. Residual learning enhanced grey-box modelling for indoor temperature prediction and IEQ assessment. Buildings, 16(5):964, 2026. https://doi.org/10.3390/buildings16050964

Elsayed, A., Ghaith, M., Yosri, A., Li, Z., and El-Dakhakhni, W. Genetic programming expressions for effluent quality prediction: Towards AI-driven monitoring and management of wastewater treatment plants. Journal of Environmental Management, 356:120510, 2024. https://doi.org/10.1016/j.jenvman.2024.120510

Farhi, N., Kohen, E., Mamane, H., and Shavitt, Y. Prediction of wastewater treatment quality using LSTM neural network. Environmental Technology & Innovation, 23:101632, 2021. https://doi.org/10.1016/j.eti.2021.101632

Henze, M., Gujer, W., Mino, T., and van Loosdrecht, M. C. M. (2000). Activated Sludge Models ASM1, ASM2, ASM2d and ASM3. IWA Scientific and Technical Report No. 9. IWA Publishing, London. https://doi.org/10.2166/9781780402369

Hewamalage, H., Ackermann, K., and Bergmeir, C. Forecast evaluation for data scientists: Common pitfalls and best practices. Data Mining and Knowledge Discovery, 37:788–832, 2022. https://doi.org/10.1007/s10618-022-00894-5

Hochreiter, S. and Schmidhuber, J. Long short-term memory. Neural Computation, 9(8):1735–1780, 1997. https://doi.org/10.1162/neco.1997.9.8.1735

Kipf, T. N. and Welling, M. (2017). Semi-supervised classification with graph convolutional networks. In Proceedings of the 5th International Conference on Learning Representations (ICLR). https://doi.org/10.48550/arXiv.1609.02907

Li, D., Yang, C., and Li, Y. A multi-subsystem collaborative Bi-LSTM-based adaptive soft sensor for global prediction of ammonia-nitrogen concentration in wastewater treatment processes. Water Research, 254:121347, 2024. https://doi.org/10.1016/j.watres.2024.121347

Li, K., Duan, H., Liu, L., Qiu, R., Van Den Akker, B., Ni, B., Chen, T., Yin, H., Yuan, Z., and Ye, L. An integrated first principal and deep learning approach for modeling nitrous oxide emissions from wastewater treatment plants. Environmental Science & Technology, 56(10):6592–6601, 2022. https://doi.org/10.1021/acs.est.1c05020

Li, Y., Kong, B., Yu, W., and Zhu, X. An attention-based CNN-LSTM method for effluent wastewater quality prediction. Applied Sciences, 13(12):7011, 2023. https://doi.org/10.3390/app13127011

Liu, X., Shi, Q., Liu, Z., and Yuan, J. Using LSTM neural network based on improved PSO and attention mechanism for predicting the effluent COD in a wastewater treatment plant. IEEE Access, 9:146082–146096, 2021. https://doi.org/10.1109/access.2021.3123225

Long, K., Sheng, Z., Shi, H., Li, X., Chen, S., and Ahn, S. Physical enhanced residual learning (PERL) framework for vehicle trajectory prediction. Communications in Transportation Research, 2025. https://doi.org/10.1016/j.commtr.2025.100166

Lv, J., Du, L., Lin, H.-Y., Wang, B., Yin, W., Song, Y., Chen, J., Yang, J., Wang, A., and Wang, H. Enhancing effluent quality prediction in wastewater treatment plants through the integration of factor analysis and machine learning. Bioresource Technology, 393:130008, 2023. https://doi.org/10.1016/j.biortech.2023.130008

Morris, M. D. Factorial sampling plans for preliminary computational experiments. Technometrics, 33(2):161–174, 1991. https://doi.org/10.1080/00401706.1991.10484804

Pisa, I., Santín, I., Morell, A., Vicario, J., and Vilanova, R. LSTM-based wastewater treatment plants operation strategies for effluent quality improvement. IEEE Access, 7:159773–159786, 2019. https://doi.org/10.1109/access.2019.2950852

Rieger, L., Gillot, S., Langergraber, G., Ohtsuki, T., Shaw, A., Takács, I., and Winkler, S. Guidelines for using activated sludge models. IWA Scientific and Technical Report, 2012. https://doi.org/10.2166/9781780401164

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., and Polosukhin, I. (2017). Attention is all you need. In Advances in Neural Information Processing Systems (NeurIPS), volume 30. Curran Associates, Inc. https://doi.org/10.48550/arXiv.1706.03762

Verhaeghe, L., Verwaeren, J., Kirim, G., Daneshgar, S., Vanrolleghem, P. A., and Torfs, E. Towards good modelling practice for parallel hybrid models for wastewater treatment processes. Water Science and Technology, 89(11):2971–2990, 2024. https://doi.org/10.2166/wst.2024.159

Wei, Q., Xu, Z., and Yin, H. Enhanced nitrogen prediction and mechanistic process analysis in high-salinity wastewater treatment using interpretable machine learning approach. Bioresource Technology, 401:132393, 2025. https://doi.org/10.1016/j.biortech.2025.132393

Wu, X., Chen, M., Zhu, T., Chen, D., and Xiong, J. Pre-training enhanced spatio-temporal graph neural network for predicting influent water quality and flow rate of wastewater treatment plant. Science of The Total Environment, 951:175411, 2024. https://doi.org/10.1016/j.scitotenv.2024.175411

Xie, Y., Chen, Y., Lian, Q., Yin, H., Peng, J., Sheng, M., and Wang, Y. Enhancing real-time prediction of effluent water quality of wastewater treatment plant based on improved feedforward neural network coupled with optimization algorithm. Water, 14(7):1053, 2022. https://doi.org/10.3390/w14071053

Xie, Y., Chen, Y., Wei, Q., and Yin, H. A hybrid deep learning approach to improve real-time effluent quality prediction in wastewater treatment plant. Water Research, 250:121092, 2023. https://doi.org/10.1016/j.watres.2023.121092

Xiong, Z., Liu, X., Igou, T. K. S., Li, Z., and Chen, Y. Using hybrid machine learning to predict wastewater effluent quality and ensure treatment plant stability. Water, 17(13):1851, 2025. https://doi.org/10.3390/w17131851

Zhang, S., Jin, Y., Chen, W., Wang, J., Wang, Y., and Ren, H. Artificial intelligence in wastewater treatment: A data-driven analysis of status and trends. Chemosphere, 336:139163, 2023. https://doi.org/10.1016/j.chemosphere.2023.139163