Unnatural inflows of industrial effluents into freshwater are a consistent source of danger to both ecology and human beings. The conventional monitoring methods are usually supported by manual sample collection and laboratory tests, which take a long time to determine disease outbreaks and cannot afford quick monitoring that, could be implemented as a measure of timely action. This paper introduces the design of a high-grade early warning system, completely designed within MATLAB, which combines a physics-based dynamic representation of dissolved oxygen (DO) depletion through an improved Streeter-Phelps model with a statistically modeled data neural network (ANN) classifier taught to determine the presence of the pollution. Fence - to-fence data was used to calibrate and validate the model, as more than 10,000 multi-parameter records of chemical oxygen demand (COD), DO, pH, and heavy metal concentrations (mass per unit volume) of compounds such as Zn, Pb, and Cd are available. Conclusions show that ANN attained predictive accuracy surpassing 92% and strongly beat other similar models, such as support vector machines (SVM) and random forests (RF). Moreover, the framework equipped with MATLAB could provide an alert within two minutes or so after the data collection point, which was a massive increase over the conventional detection strategies. Such results support the opportunities of hybrid simulation-intelligence systems regarding proactive water quality control, providing regulatory organizations and industrialists with powerful instruments to reduce pollution outcomes.
Nooruldeen et al. (Mon,) studied this question.