Gated Recurrent Unit with Self Attention Mechanism for Optimizing Wastewater Treatment Process

The Wastewater Treatment Process (WWTP) is utilized to avoid water source pollution which is reused for agricultural and enhances the sanitation conditions. The main aim of WWTP is eliminate the nutrients like nitrogen that occurs in ammonia form in waste. Artificial intelligence technology has been widely applied in the sewage treatment process, especially the application of some intelligent algorithms to improve the efficiency and accuracy of sewage treatment. The Gated Recurrent Unit with Self-Attention Mechanism (GRU-SAM) is proposed in this work which is used to enhance the WWTP by using climate dataset. The dataset is preprocessed by min-max normalization which is used to reduce the relative connection among quantities and remove the dimension influence among indicators. Then, Discrete Wavelet Transform (DWT) is used for feature extraction and GRU-SAM is used for classification. The performance is estimated by Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The GRU-SAM attains MSE 0.005, RMSE 0.0647, MAE 0.009 and MAPE 16.58 which is better compared to existing techniques such as Deep Learning-based Bidirectional Long Short-Term Memory-Water Quality Assessment (DLBL-WQA) and Artificial Neural Network-WT-Long Short-Term Memory (ANN-WT-LSTM).