Mutual information-kernel principal component analysis-Elman network-based medium-long-term runoff forecast method

Abstract

The invention discloses a mutual information, kernel principal component analysis and Elman network-based medium-long-term runoff forecast method. The method comprises the following steps of collecting meteorological hydrological data and establishing a one-to-one correspondence relationship of an index time sequence and a runoff time sequence; selecting out indexes with high remarkable and standard mutual information by adopting a standard mutual information method; extracting a principal component of screened index data by using a kernel principal component analysis method; building an Elman neutral network model; after the principal component is standardized by z-score, generating a training sample to perform supervised training on a network, generating a check sample to perform check on the network, and calculating each evaluation index value; and repeating one-time forecast for multiple times, and taking a set average of multiple forecasts as a final forecast value. Linear and nonlinear relationships between the meteorological hydrological data and runoff can be fully mined; and a mathematic relationship model is built, so that more accurate and stable forecast of medium-long-term runoff volume is realized.

Description

technical field [0001] The invention belongs to the field of information technology, in particular to a medium and long-term runoff forecasting method based on mutual information-kernel principal component analysis-Elman network. Background technique [0002] Accurate medium and long-term runoff forecasting is an important prerequisite for guiding the comprehensive development and utilization of water resources, scientific management and optimal dispatch. [0003] At present, the commonly used medium and long-term runoff forecasting methods are based on statistical methods, that is, forecasting is realized by finding the statistical relationship between forecasting objects and forecasting factors. The key issues in applying statistical methods to medium and long-term runoff forecasting include the following three aspects: [0004] (1) Preliminary selection of predictors: For the primary selection of predictors, the commonly used method is linear correlation analysis (such a...

AI Technical Summary

Problems solved by technology

This method ignores the relationship between the data higher than the second order, so the extracted principal components are not optimal;

[0010] (3) The model used to establish the optimal mathematical relationship between the forecast object and the predictor. The commonly used multiple regression is actually a kind of linear fitting, which cannot reflect the nonlinear relationship between the forecast object and the pr

Images