Advancing Environmental Sustainability through Implementation of Artificial Neural Networks for Wastewater Treatment Model Prediction and Remediation

Abstract

This research examines the effectiveness of various artificial neural network technologies in enhancing wastewater treatment models prediction and remediation on the standpoint of environmental sustainability. This study was done via ten experiments that delve into the performance of the four types of ANN architectures known as Feedforward Neural Networks, Recurrent Neural Networks, Long Short-Term Memory Networks, and Gated Recurrent Unit Networks. This process model’s validation findings affirm the models’ capability to forecast the outcome of the intervention and guide the correct form of remediation. FNNs yield consistently low levels of MSE: between 0.010 and 0.020, while the R^ 2 is high and stays between 0.910 and 0.960. Meanwhile, RNNs, LSTM Networks, and GRU Networks include slightly higher MSE values but feature a stronger correlation with R^2 : between 0.880 and 0.940. These results indicate that the ANN technologies have the potential to model the treatment processes. They can be utilized to ensure more optimal treatment processes and reduce environmental contamination.