Research on Monthly Precipitation Prediction in Lanzhou City Based on Machine Learning Model
Accurate forecasting of monthly precipitation is of great significance for national production as well as disaster prevention and mitigation.However,it is difficult for a single model to complete the task of accurate precipitation prediction.We combine CEEMDAN with the error reverse communication model(BP)and with the long short-term memory neural network(LSTM),respectively.And we compare Lanzhou's precipitation data with the precipitation predictions with a single LSTM model,ARIMA model and BP model.The research results show that the two composite models effectively improve the fitting of observation values and prediction values.Thus,the problem of low accuracy in peak prediction is overcomed,showing significantly higher performance than the comparative models.
万方数据
维普
