Transformer-BiLSTM-based water level prediction model

Underground water level monitoring and prediction are increasingly becoming the key means to deal with urban waterlogging. Traditional prediction methods mainly rely on a single statistical model or shallow machine learning algorithm. These methods are difficult to effectively capture the impact of complex underground water level data and meteorological environment data on water level trend. To solve the above problems, a hybrid model of water level prediction based on transformer and bilstm is proposed. The model combines the transformer's self attention mechanism and bilstm's two-way information processing ability, and can effectively deal with the long-distance dependence and complex time patterns in water level time series data. The test of water level data and meteorological data collected in the field shows that compared with the traditional model, the proposed model can predict the underground water level change more accurately.