Multi-source Heterogeneous Data Fusion for Toxin Level Quantification

Abstract The operational management of wastewater treatment plants (WWTP) is a complex activity due to the biological phenomena’ intricate nature. This complexity hinders the adoption of first principles approaches, which lack the necessary accuracy to be adopted in practice. Data-driven methodologies also face significant challenges in processing the different information sources available. In this work, we present a data-driven and model-agnostic data-fusion framework to estimate the concentration level of a toxin in the effluent, using heterogeneous data (sensor data, images, laboratory measurements) collected at different locations in the process. Single- and multi-source modeling approaches are applied and compared. Among the methodologies tested, Bayesian fusion stands out as presenting a good balance in terms of accuracy, stability, and flexibility.