Hybrid Water Quality Prediction with Bidirectional Long Short-Term Memory and Encoder-Decoder

Accurate and real-time prediction of water quality not only helps to assess the environmental quality of water, but also effectively prevents and controls water quality emergencies. In recent years, neural networks represented by Bidirectional Long Short-Term Memory (BiLSTM) and Encoder-Decoder (ED) frameworks have been shown to be suitable for prediction of time series data. However, traditional statistical methods cannot capture nonlinear characteristics of the water quality, and deep learning models often suffer from gradient disappearance and gradient explosion problems. This work proposes a hybrid water quality prediction method called VBEG, which combines V ariational Mode Decomposition (VMD), B iLSTM, an E D structure, and G enetic Simulated annealing-based particle swarm optimization (GSPSO). VBEG first adopts VMD to deal with nonlinear features in the original time series. Then, VBEG combines BiLSTM and the ED structure to capture bi-directional long-term correlations, and realize dimensionality reduction, respectively. Furthermore, VBEG adopts GSPSO to optimize its hyperparameters. Experimental results with real-life datasets demonstrate that the proposed VBEG outperforms two current state-of-the-art algorithms in terms of prediction accuracy.