Water flow prediction system and method based on machine learning

Abstract

The invention discloses a water flow prediction system based on machine learning. The water flow prediction system comprises N station ends and a prediction end, each station end sends single-end original rainfall data to the prediction end, and the station end corresponding to the target hydrological station further sends original flow data; the prediction end also obtains prior rainfall data ofeach hydrological station in the future P hours and obtains original rainfall data from all the obtained single-end original rainfall data, preprocesses the original rainfall data and the original flow data, constructs input data based on the processed rainfall data, the processed flow data and the prior rainfall data, inputs the data into a flood flow prediction model to carry out loop iterationprediction, extracts time sequence characteristics from the input data of the iteration and carries out classification prediction during each iteration to obtain a flow prediction value of the targethydrological station in the next hour, and reconstructs the input data aiming at the iteration in the next hour in the future until P iterations are completed, and and obtains a flow prediction valueof the target hydrological station in the future P hours.

Description

technical field [0001] The invention belongs to the field of flood forecasting, and in particular relates to a water flow forecasting system and method based on machine learning. Background technique [0002] Floods are one of the common natural disasters. Hundreds of millions of people are affected by floods and displaced every year, and the financial and material losses caused by floods are also huge. Effectively predicting flood flow and issuing early warnings is of great significance for flood control and disaster reduction. [0003] The current flood flow forecasting models are mainly divided into traditional physical models and intelligent flood forecasting models. Traditional physical models, such as the Xin'anjiang model, calculate the parameters of the physical process on the premise of fully mining the physical characteristics of the local topography, evaporation, and vegetation coverage, and finally formulate a set of regionally specific predictions Model. The ...

AI Technical Summary

Problems solved by technology

[0004] However, most of the existing flood flow forecasting models belong to single-point forecasting, that is, predicting the flow situation at a point in time in the future. In actual situations, the flow data obtained at a single point in time from the prediction lacks practical application value.

Images