Research on water quality prediction based on PE-CNN-GRU hybrid model

Sewage treatment is a complex and nonlinear process. In this paper, prediction method based on convolutional neural network (CNN) gated recurrent unit (GRU) hybrid proposed for the of dissolved oxygen concentration in sewage treatment. Firstly, akima 's used to complete filling preprocessing missing data, then integrated empirical mode decomposition (EEMD) algorithm denoise key factors water quality data. Pearson correlation analysis select better parameters as input model. Then, CNN convolve data sequence extract feature components CNN-GRU prediction, predicted output value obtained. The mean absolute error (MAE), root square (RMSE) (MSE) were evaluation criteria analyze results By comparing with RNN model, LSTM GRU model CNN-LSTM show that PCA-EEMD-CNN-GRU (PE-CNN-GRU) paper has significantly improved accuracy concentration.