Sludge Age Prediction Using Machine Learning (Deep Neural Network vs Deep Bi-LSTM Model)

The Sludge Age (SAGE) is a significant operational parameter in activated sludge processes. However, precisely estimating it can be difficult due to non-linear data and changeable operating conditions. The objective of this project is to create a bidirectional long short-term memory (Bi-LSTM) network to forecast sludge age (SAGE) using intricate time-series data obtained from a Wastewater Treatment Plant (WWTP). The Bi-LSTM architecture is recognized for efficiently handling time series and non-uniform data. The datasets used for training and evaluating the Bi-LSTM model was gathered from the Nine Springs Wastewater Treatment Plant located near Madison, Wisconsin, United States. Following the utilization of historical data for training, the suggested Bi-LSTM model outperform the DNN model and successfully forecasted SAGE, offering potential advantages to WWTP managers in enhancing operational performance, system management, and process stability.