Digital water: artificial intelligence and soft computing applications for drinking water quality assessment

Water quality deterioration in drinking water systems (i.e., system failure) causing serious outbreaks have frequently been happening around the world. These failures can be predicted through a real-time drinking water quality monitoring system for timely actions. Although Supervisory Control and Data Acquisition (SCADA) has been commonly used for this purpose, this system suffers considerable limitations, such as the scalability of sensors, lack of predictive ability, and increased burden on operators overwhelmed by superfluous notifications. Proficient artificial intelligence & soft computing (AI & SC) techniques and cloud Internet of Things (IoT) may significantly reduce the reliance on operators and eventually improve system operations. This study critically reviewed the literature published from 2000 to 2020 to evaluate AI & SC applications’ trends in drinking water quality management while developing a roadmap for autonomous digital water quality management. The investigation reveals that AI & SC were primarily used for assessing drinking and surface water quality. These techniques were largely applied to effectively predict, evaluate, and control water quality. AI & SC were also used to model physicochemical (basic and complex) and microbiological parameters. The most commonly applied AI & SC techniques are the multilayer perceptron-based artificial neural network, general regression neural network, support vector machine, Bayesian networks, and the adaptive neuro-fuzzy inference system. Additional articles published after 2020 were also reviewed to compare with the measures proposed in the roadmap. This roadmap guides water utilities, policy-makers, and regulators to achieve reliable water quality in a smart and efficient manner. Graphical abstract