Year: 2026 | Month: February | Volume: 13 | Issue: 2 | Pages: 381-389
DOI: https://doi.org/10.52403/ijrr.20260236
Hybrid Data-Driven Models and Intelligent Control Strategies for Advanced Wastewater Treatment: Integrating Modular Learning, Fault-Tolerant Systems, and Energy Optimization
Hendri Iyabu1, Hasim2, Fitryane Lihawa3, Weny J.A Musa4, Novri Youla Kandowangko5, Marike Mahmud6
1Doctoral Program in Environmental Science, Universitas Negeri Gorontalo, Gorontalo, Indonesia
2,3,4,5,6Postgraduate Program, Universitas Negeri Gorontalo, Gorontalo, Indonesia
Corresponding Author: Hendri Iyabu
ABSTRACT
Modern wastewater treatment plants (WWTPs) face the dual challenge of maintaining high treatment efficiency while adhering to increasingly stringent environmental regulations and minimizing operational costs. In response, the field of process control has evolved, shifting from traditional methods to more sophisticated, hybrid data-driven models and intelligent control strategies. This review synthesizes the most recent literature on advanced WWTP control, focusing on four major advancements: (1) system-level predictive models based on modular learning architectures, (2) adaptive and reinforcement learning–based optimal control strategies, (3) fault-tolerant systems for resilience under actuator and sensor failures, and (4) energy-efficient aeration and multi-objective optimization techniques. Recent developments in Modular Merged LSTM (MM-LSTM), hybrid adaptive critic designs, fuzzy-neural control models, and digital twin frameworks provide enhanced prediction, real-time adaptability, and robust performance even under operational disturbances. Energy optimization strategies using heuristic initialization and reinforcement learning have demonstrated up to 25% reductions in energy consumption for aeration systems. Additionally, the integration of fault estimation and self-healing control mechanisms has significantly improved system robustness under sensor and actuator faults. Through detailed comparative tables, this review consolidates the current state of research and outlines key trends, challenges, and gaps in deploying these models in large-scale WWTPs. Future research directions focus on scaling hybrid models, improving transfer learning methods, and developing digital twin standards for real-world implementation.
Keywords: Hybrid Data-Driven Models, Intelligent Control Strategies, Predictive Models, Energy-Efficient Aeration, Digital Twin Frameworks
[PDF Full Text]