Research Of Water Quality Prediction Method Based On Attention Mechanism And Long Short Term Memory Neural Network

Water quality prediction,which is of great significance to water pollution prevention,is a basic work in water resources protection.It establishes a water quality prediction model based on the historical water quality data to predict the changing trend of water quality data in the future.Water quality prediction is an important means to transform water pollution from post-event treatment to pre-event prevention,and provides a basis for the formulation of water pollution prevention scheme.Therefore,accurate water quality prediction has important research significance.Water quality data is affected by time,physics and other factors,and it has complex characteristics such as temporality,diversity and chaos.At present,the traditional water quality prediction methods fail to fully consider the impact of these characteristics on the prediction results,which causes the problem of low prediction accuracy.Therefore,in order to fully consider the temporality,diversity and chaos of water quality data and improve the accuracy of water quality prediction,this thesis studies the water quality prediction method based on attention mechanism and long short term memory neural network.The specific work contents are as follows:In view of the temporality and diversity of water quality data,this thesis proposes a water quality prediction method based on SA-BiLSTM.First,the self attention(SA)mechanism is introduced into the bidirectional long short term memory(BiLSTM)neural network to establish a water quality prediction model based on SA-BiLSTM.Second,BiLSTM is used to process the temporality of water quality data at different times to mine the time sequence information of water quality data,so as to make full use of the temporality.At the same time,SA is used to calculate the similarity weight of water quality data at different times to capture the important information of water quality data,so as to make full use of the diversity.Third,this thesis summarizes the process of water quality prediction method based on SA-BiLSTM,and standardizes the specific steps of prediction.Finally,the proposed method is applied to two practical water quality data sets: Taihu Lake and Victoria Lake.Experimental results show that the proposed method has better performance than the existing methods.In view of the chaos of water quality data,this thesis proposes a water quality prediction method based on t-SNE and SA-BiLSTM.First,the t-distributed stochastic neighbor embedding(t-SNE)is used to effectively extract the features of water quality data,and the extracted water quality features are used as the input of prediction model to reduce the adverse impact caused by the chaos of water quality data.Second,this thesis summarizes the process of water quality prediction method based on t-SNE and SA-BiLSTM,and standardizes the specific steps of prediction.Finally,the proposed method is also applied to the two practical water quality data sets.Experimental results show that the proposed method can further improve the accuracy of water quality prediction after introducing t-SNE into SA-BiLSTM.