Insights into the application of explainable artificial intelligence for biological wastewater treatment plants: Updates and perspectives
Abdul Gaffar Sheik, Arvind Kumar, Chandra Sainadh Srungavarapu, Mohammad Azari, Seshagiri Rao Ambati, Faizal Bux, Ameer Khan Patan
Abstract
Explainable artificial intelligence (XAI) is an interactive platform that assists users in comprehending the decisions and predictions made by machine learning (ML) models. This allows users to enhance their knowledge of ML models and their functioning, which not only helps in mitigating bias and errors but also aids in improving user decision-making confidence. XAI, due to its ability to increase the model output interpretation, has gained significant attention in biological wastewater treatment plants (WWTPs). This is owing, in particular, to the fact that it facilitates the experts in steering knowledge about the predictions and decisions made by ML, thus guaranteeing that the model decisions are fair and unbiased. ML has made amazing advances in recent years, thanks to its exponential growth in possessing the power to process massive volumes of data, allowing it to be widely embraced in WWTPs. This review seeks to illustrate the potential of XAI for WWTP applications such as process modeling and control, soft sensing, fusion of data, and the internet of things, and fill the knowledge gap by thoroughly introducing XAI techniques and their use in smart wastewater engineering. Overall, the features of XAI can aid in establishing reliable and efficient water resource management , which is quintessential to achieving environmental sustainability . It is envisioned that the prospects offered would spark new lines of study, helping to reduce the current skepticism and apprehension about ML adoption and integration in WWTP.