A Neural Network Method for Predicting River Flow Based on Multi-section Water Level

Abstract

The invention relates to a neural network method for predicting river flow based on multi-section water levels, the steps of which are: obtaining the flow data and water level data of the previous year known by the hydrological station to be sought, and the water level data of m water level stations upstream of the hydrological station to be sought and the water level data of n water level stations downstream of the desired hydrological station, and number the water level data to obtain the water level data of each water level station; The water level data and the water level data of the adjacent water level stations downstream of the desired hydrological station all constitute a set of independent variables H i ; Each set of independent variables H i and its corresponding time flow data Q i Perform BP neural network fitting to obtain the BP neural network model; according to the known traffic data Q m The corresponding time, the water level data measured at this time is used as the independent variable H m ; Set the argument H m Substitute into the BP neural network model to get the corresponding flow data Q m , and then get the water level-flow relationship curve to realize the prediction of river flow.

Description

technical field [0001] The invention relates to a method for predicting river flow, in particular to a neural network method for predicting river flow based on multi-section water levels. Background technique [0002] At present, the flow data of natural rivers are important data in the research process of many disciplines such as hydrology and river dynamics. However, in the past engineering experience, it is very difficult to directly measure the flow of the actual river, and it is time-consuming, costly, and has great limitations. In contrast, water level data measurement is simple and easy, with complete data. If the relationship between river discharge and water level can be obtained from the existing water level data and flow data, and the water level-flow curve can be drawn, then the flow data can be obtained from the water level data, which greatly improves the efficiency of scientific research and engineering. [0003] In the previous water level-discharge curve m...

AI Technical Summary

Problems solved by technology

[0009] With the gradual development of water energy, the water storage and discharge process of the reservoir has had an important impact on the water level-flow relationship of each hydrological station, especially the hydrological station located at the end of the reservoir. During the storage period: the water level of the reservoir rises, The water level of the section also increases correspondingly, but the flow rate is still low, and the original water level-flow curve is no longer applicable

Taking the year as the time scale, due to the impact of reservoir regulation and storage, the flow of the corresponding river section does not meet the condition of constant flow, and the hydraulic gradient J is quite different at different time periods throughout the year, resulting in a gap between water level and flow. A corresponding relationship is destroyed, and the original water level-flow curve is no longer applicable

However, the relationship between water level and flow in actual rivers is very complicated, especially in river sections affected by water conservancy and hydropower projects, often appearing in the shape of multi-socket curves, and the water level and flow of a single section of a river show obvious non-monotonicity The accuracy of single-value curve fitting is very poor, which affects the progress of scientific research and engineering

Images