A Water Quality Prediction Method Based on Long Short-Term Memory Neural Network Optimized by Cuckoo Search Algorithm

Water quality prediction is of positive significance for the protection of water sources. In order to grasp the future water quality of the reservoir, the measured data of the four indicators of pH, biochemical oxygen demand of five days(BOD5), ammonia nitrogen(NH
3
-N)and dissolved oxygen(DO)from 2017 to 2021 were selected as training samples. The Long Short-Term Memory (LSTM) neural network was optimized by using the Cuckoo Search(CS)algorithm to predict the four indicators of the reservoir in 2022. LSTM neural network model and BP fully connected neural network model are established and compared with them. The experimental results show that the mean absolute error and root mean square error of the prediction model based on CS-LSTM are lower than those of the comparison model, and the coefficient of determination is higher than that of the comparison model, which is better than the LSTM model.