River flood prediction method based on similarity matching and extreme learning machines

Abstract

The invention relates to a river flood prediction method based on similarity matching and extreme learning machines. The method comprises the following steps of firstly, obtaining an optimal flow prediction model respectively corresponding to each history flood sample based on extreme learning machine models; secondly, selecting front k preset history flood samples as reference flood samples based on a similarity matching value from big to small; and lastly, carrying out flow value prediction based on real-time water flow features and real-time rainfall capacity features and carrying out flood judgment according to the obtained flow prediction value. According to the river flood prediction method based on similarity matching and the extreme learning machines, the defects of the existing technology can be overcome and the actual prediction accuracy of the river flood can be effectively improved.

Description

technical field [0001] The invention relates to a river flood prediction method based on similarity matching and extreme learning machines, and belongs to the technical field of river monitoring. Background technique [0002] Flood forecasting directly serves the national economic construction and is an important non-engineering measure for disaster prevention and mitigation. Due to the large number of small and medium-sized rivers, complex terrain, diverse climate types, and various vegetation types in China, and the generally low accuracy of rainfall forecasts, flood forecasts are not accurate enough. Natural disasters such as floods and mudslides caused by short-term heavy rainfall every year cause huge losses of life and property to human beings, so it is very important to improve the accuracy of flood forecasting. [0003] The data-driven flood forecasting model is a kind of black box method that does not consider the physical mechanism of the hydrological process, and...

AI Technical Summary

Problems solved by technology

However, the artificial neural network needs to set a large number of parameters during the training process, and the debugging of the parameters is also very complicated, and it is easy to fall into the local optimal solution.

Literature [Wang Lina, Li Yuean, Chen Xiaohong. Rainfall-runoff forecasting research based on support vector machine [J]. Hydrology, 2009, 29(1): 13-16] Using support vectors to build a rainfall-runoff forecasting model, compared with artificial neural The network has achieved high prediction accuracy, but because the historical flood data contains a variety of samples with different data distribution characteristics, a single model cannot describe the characteristics of all data well at the same time

Images