Directed graph deep neural network runoff probability forecasting method, system, equipment and terminal

Abstract

The invention belongs to the technical field of hydrology and water resources, and discloses a directed graph deep neural network runoff probability forecasting method, system, device and terminal, the directed graph deep neural network runoff probability forecasting method comprises the following steps: constructing a hydrological station and meteorological station directed graph structure; according to the multi-site directed graph, establishing a directed graph deep neural network forecasting model by combining two processes of a spatial information capturing process and a feature aggregation process; constructing a data set consisting of forecast runoff and forecast factors thereof, and training a directed graph deep neural network by adopting an Adam optimization algorithm to obtain a high-precision multi-step runoff forecast result; reconstructing a hidden Markov regression training data set according to the forecast result obtained by the directed graph deep neural network and the actually measured runoff value; and training the hidden Markov regression model to obtain a runoff probability forecast result. According to the method, the graph theory and the neural network model are combined, the spatial information is captured through convolution operation, the rainfall runoff process is aggregated through the multi-layer network, and the runoff forecasting precision is improved.

Description

technical field [0001] The invention belongs to the technical field of hydrology and water resources, and in particular relates to a directed graph deep neural network runoff probability forecasting method, system, equipment and terminal. Background technique [0002] At present, accurate and reliable runoff forecasting can provide useful data basis for decision-making of reservoir group scheduling, and is of great significance for increasing power generation benefits, reducing flood control risks, and improving the benefits of comprehensive utilization of water resources. With the development of computer technology, machine learning and deep learning methods such as support vector machine regression, neural network, and long-term short-term memory network have shown excellent performance in runoff forecasting. However, this kind of black-box model based on statistics often has two disadvantages: one is that the black-box model is only the transmission between data and its m...

AI Technical Summary

Problems solved by technology

[0008] Aiming at the problems existing in the prior art, the present invention provides a directed graph deep neural network runoff probability forecasting method, system, equipment and terminal, especially relates to a directed graph deep neural network runoff based on a directed graph deep neural network The probabilistic forecasting method, system, equipment and terminal aim to solve the technical problems that the existing statistical learning runoff forecasting methods have no specific hydrophysical meaning and cannot quantify the uncertainty in the forecasting

Images