Intelligent set evaluation method and system for basin water and sediment research model

Abstract

The invention discloses an intelligent set evaluation method and system for a basin water and sediment research model, and the method comprises the following steps: (1), determining a research objectwhich comprises a basin, dependent variables, independent variables and a time scale; (2) screening an optimal model; randomly scrambling the determined watershed water and sediment data set and thensegmenting the watershed water and sediment data set into a training set and a test set; selecting different machine learning methods, dividing multiple groups under different time scales and different independent variable combinations to cover all possible conditions, obtaining a group of optimal models through parameter obtaining and screening in each condition, and screening out a group of optimal models with the best result through each method to serve as final optimal models; and (3) evaluating the final optimal model based on three different index systems, wherein the index systems comprise a dimensionless index and a dimensionless index for evaluating the excellence of the model and an evaluation index based on a minimum information criterion for balancing the excellence and the complexity of a model fitting result. And a quantitative evaluation result is provided for the applicability of each type of models of the set by using a unified standard.

Description

Technical field [0001] The invention belongs to a collective assessment technical system, and specifically relates to an intelligent collective assessment method and system for a water and sediment research model in a river basin. Background technique [0002] Since the 21st century, the water and sand regime of the Yellow River has undergone unprecedented drastic changes. The in-depth analysis of the Yellow River's water and sediment regime is of great significance. It is related to the direction of the management of the wide section of the lower Yellow River, the allocation and utilization strategy of the Yellow River's water resources, the layout of water and sediment control projects and the establishment of an overall Yellow River governance strategy. Commonly used empirical models and process-based physical models, etc., can show that the Yellow River water and sediment changes are the result of the comprehensive effects of human activities and climate change, but the resul...

AI Technical Summary

Problems solved by technology

[0003] In addition, the existing methods have certain limitations. The disadvantage of the empirical model method is: for the sake of simplicity of calculation, the empirical model usually only selects the precipitation P or other forms of precipitation that have the greatest correlation with the output (runoff or sediment transport) As an input, other factors are neglected, resulting in insufficient interpretation of the input data; the research period is divided into the base period and the period affected by human activities, that is, it is assumed that the watershed is not affected by human activities during the base period, and the basin is not affected by human activities during the period affected by human activities. , the river basin is affected by the same degree of human activities every year, but in fact the runoff or sediment transport in the river basin is affected by human activities in different degrees in different years; According to the principle, the calculated value of runoff or sediment transport during the period affected by human activities is calculated, and the difference between the calculated value and the actual value is interpreted as the error caused by human activities, but in fact it also includes the error caused by the model structure. The contribution rate of human activities calculated by the empirical model is not accurate

If you choose a fixed rate period, the obtained model parameters are the optimal parameters in the rate period, but it cannot be guaranteed to be the optimal parameter in the simulation period, which may cause overfitting

Therefore, when using the parameters obtained regularly by the rate to simulate the future situation, it may cause a large error

Images