Hydropower station power generation tail water dissolved oxygen prediction method and device

The invention provides a hydropower station power generation tail water dissolved oxygen prediction method and device. The method comprises the steps that a dissolved oxygen prediction value sequence of real-time data prediction of an LSTM-DNN prediction model constructed by historical data of dispatching operation of a water quality monitoring station, a weather station and a hydropower station is obtained; calculating a deviation sequence according to a dissolved oxygen monitoring value sequence and a dissolved oxygen prediction value sequence in the historical data, and training the generalized additive model according to distribution parameters of the deviation sequence to obtain a dissolved oxygen prediction deviation distribution GAMLSS model; inputting a real-time temperature comprehensive factor calculated by the water temperature and the air temperature in the real-time data into a dissolved oxygen prediction deviation distribution GAMLSS model to obtain a confidence interval of the dissolved oxygen