Advancements in machine learning modelling for energy and emissions optimization in wastewater treatment plants: A systematic review

Abstract Wastewater treatment plants (WWTPs) are high‐energy consumers and major Greenhouse Gas (GHG) emitters. This review offers a comprehensive global overview of the current utilization of machine learning (ML) to optimize energy usage and reduce emissions in WWTPs. It compiles and analyses findings from over a hundred studies primarily conducted within the last decade. These studies are organized into five primary areas: energy consumption (EC), aeration energy (AE), pumping energy (PE), sludge treatment energy (STE) and greenhouse gas (GHG). Additionally, they are further categorized based on learning type, the scale of application, geographic location, year, performance metrics, software, etc. ANNs emerged as the most prevalent, closely trailed by FL and RF. While GA and PSO are the predominant metaheuristic approaches. Despite increasing complexity, researchers are inclined towards employing hybrid models to enhance performance. Reported reductions in energy consumption or GHG emissions spanned various ranges, falling within the 0–10%, 10–20% and >20% brackets.