A Multi-step Forecasting Method for Chaotic Time Series Based on Density Peak Clustering

Abstract

The invention discloses a multi-step prediction method for chaotic time series based on density peak clustering. In the method, adjacent orbits are selected multiple times, and the phase point boundary is determined through a clustering algorithm. Re-select the nearest neighbor orbit in the form of points, from the existing one-time acquisition of adjacent orbits based on the target phase point to multiple acquisition of adjacent orbits, combined with the clustering algorithm, without artificially determining the cluster center, dynamically assigning the boundary of the phase point, reducing The influence of human factors has improved the problem of the multi-step prediction accuracy caused by the one-time acquisition of the existing model. The experiment shows that the method of the present invention has better generalization performance than the existing multi-step prediction filter model. The impact is smaller and the training set required is smaller.

Description

technical field [0001] The invention relates to a time series prediction method, in particular to a multi-step prediction method of a chaotic time series based on density peak clustering. Background technique [0002] Chaos is a seemingly random motion produced by a definite nonlinear dynamical system, which is sensitive to initial conditions and difficult to predict in the long run. Chaotic time series prediction has broad application prospects. Based on the Takens phase space reconstruction theory, many scholars have proposed a variety of chaotic time series prediction models. The Volterra filter has attracted extensive attention from domestic and foreign scholars for its advantages of fast training speed, strong nonlinear approximation ability, and high prediction accuracy. After a series of studies on the multi-step prediction of high-dimensional chaotic time series using the Volterra filter, the Volterra filter solves the kernel function through system identification, a...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a multi-step forecasting method for chaotic time series based on density peak clustering, which is used to improve the existing forecasting model's problems of excessive dependence on the forecasting starting point and low reliability, and is used in weather convection models etc. research process

Images