Deep Learning Models for Predicting Effluent Quality Under Variable Industrial Load Conditions
DOI:
https://doi.org/10.15662/IJRAI.2021.0405004Keywords:
Deep learning, Effluent quality prediction, Industrial wastewater, Variable load conditions, LSTM networks, Convolutional neural networks (CNN), Hybrid CNN–LSTM models, Time-series forecasting, COD and BOD prediction, Feature engineering, AI-enabled wastewater management, Predictive complianceAbstract
Industrial wastewater treatment plants (WWTPs) experience considerable variability in influent characteristics due to fluctuating industrial production schedules, seasonal shifts, equipment performance, cleaning cycles, and episodic high-strength discharges. These fluctuations challenge biological and physicochemical treatment units, often causing spikes in effluent Chemical Oxygen Demand (COD) and Biochemical Oxygen Demand (BOD), which are tightly regulated under national and international discharge standards. This study presents a comprehensive machine learning (ML) and deep learning framework—comprising Long Short-Term Memory (LSTM) networks, Convolutional Neural Networks (CNNs), and hybrid CNN–LSTM architectures to predict effluent quality under dynamic load conditions. Synthetic datasets mimicking real industrial variability were generated to evaluate model behavior. Feature engineering techniques such as rolling statistics, lag variables, and hydraulic load normalization were incorporated to strengthen predictive accuracy. Results show that hybrid CNN–LSTM models outperform traditional regression and basic ML models by capturing both temporal dependencies and transient load shocks. The findings demonstrate significant potential for AI-driven predictive effluent management, enabling real-time decision support, proactive compliance, and optimized chemical and energy usage.
References
[1] T.Y. Pai, S.H. Chuang, H.H. Ho, L.F. Yu, H.C. Su, H.C. Hu, “Predicting performance of grey and neural network in industrial effluent using online monitoring parameters”, Process Biochemistry, Volume 43, Issue 2, 2008, 199-205, https://doi.org/10.1016/j.procbio.2007.10.003
[2] Korostynska, O., Mason, A., Al-Shamma’a, A.I. (2013). Monitoring Pollutants in Wastewater: Traditional Lab Based versus Modern Real-Time Approaches. In: Mukhopadhyay, S., Mason, A. (eds) Smart Sensors for Real-Time Water Quality Monitoring. Smart Sensors, Measurement and Instrumentation, Volume 4, Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-37006-9_1
[3] T. Cheng, F. Harrou, F. Kadri, Y. Sun and T. Leiknes, "Forecasting of Wastewater Treatment Plant Key Features Using Deep Learning-Based Models: A Case Study," IEEE Access, Volume 8, 184475-184485, 2020, doi: 10.1109/ACCESS.2020.3030820.
[4] Willmott, C. J., & Matsuura, K. (2006) “On the use of dimensioned measures of error to evaluate the performance of spatial interpolators”, International Journal of Geographical Information Science, Volume 20(1), 89–102. https://doi.org/10.1080/13658810500286976
[5] Scott A. Dellana, David West, “Predictive modeling for wastewater applications: Linear and nonlinear approaches”, Environmental Modelling & Software, Volume 24, Issue 1, 2009, 96-106, https://doi.org/10.1016/j.envsoft.2008.06.002.
[6] Mamandipoor, B., Majd, M., Sheikhalishahi, S. , “Monitoring and detecting faults in wastewater treatment plants using deep learning”, Environmental Monitoring and Assessment, Volume 192, 148 (2020). https://doi.org/10.1007/s10661-020-8064-1
[7] I. Pisa, I. Santín, A. Morell, J. L. Vicario and R. Vilanova, "LSTM-Based Wastewater Treatment Plants Operation Strategies for Effluent Quality Improvement," IEEE Access, Volume 7, 159773-159786, 2019, DOI: 10.1109/ACCESS.2019.2950852.
[8] Pisa I, Morell A, Vicario JL, Vilanova R, “Denoising Autoencoders and LSTM-Based Artificial Neural Networks Data Processing for Its Application to Internal Model Control in Industrial Environments—The Wastewater Treatment Plant Control Case”, Sensors, 2020; Volume 20 (13):3743. https://doi.org/10.3390/s20133743
