An Intelligent Framework for Water Quality Prediction Using Dense Dilated Gated Recurrent Unit

Water quality is an essential thing for the wellbeing of people, plants, animals, industries, and the overall environment. Modeling Water Quality Parameters (WQP) is essential for analyzing aquatic systems. Traditional modeling approaches often involve time-consuming procedures and their performance on different datasets varies and results in less generalizability. Predicting water quality is critical for aquaculture, and environmental monitoring. However, the nonlinear dynamics affect the traditional prediction methods to provide accurate results. To overcome various challenges in water quality prediction, this work develops an intelligent mechanism with deep feature learning for high-achieving accurate prediction. Initially, the data sources are gathered from standard forums and are subjected to processing. The collected data is sent to the Restricted Boltzmann Machine (RBM) network to gather the important features. The resulting RBM features are given to the developed Dense Dilated Gated Recurrent Unit (Dense-DiGRU) model for training. The proposed Dense-DiGRU model predicts the water quality from the features and produces highly accurate prediction results. The efficacy of the introduced Dense-DiGRU model is compared with State-of-the-art techniques.