Sun-spot dynamic model and tractive predicting method based on particle filter technology

Abstract

A sunspot dynamic model based on particle filtering technology is prepared as applying dynamic parameter mechanism, using historical data of sunspot as example, comparing the fitting property of model with set domain value and dynamically regulating model parameters when property index is less than set domain value. The said particle filtering technology is analyzing method of nonlinear time sequence.

Description

technical field [0001] The invention relates to the establishment of a dynamic model in the field of signal processing and a tracking prediction method oriented to the dynamic model. In particular, it involves the establishment of a dynamic model that conforms to the law of sunspot variation, and the tracking and prediction method based on particle filter technology used in the analysis and processing of the model. Background technique [0002] The existence of the ionosphere above the earth makes it possible for human beings to have long-distance wireless communication. Solar activity is a major factor affecting the state of the ionosphere. Strong solar activity will not only cause irresistible ionospheric disturbances, but also affect or even interrupt radio communication. It will cause a strong turmoil in the earth's magnetic field—a magnetic storm, which will seriously interfere with the navigation capabilities of aircraft and ships. The earth current generated during th...

AI Technical Summary

Problems solved by technology

The first two filters do not take into account all the statistical properties of the data, often leading to poor results

The grid filter can obtain accurate results through a certain numerical integration method, but it is difficult to implement, and it will face a computational bottleneck in high-dimensional situations

At the same time, the grid filter must predefine the state space and bisect the state space, resulting in that for high probability density regions, relatively higher refinement cannot be given unless prior knowledge is applied

Hybrid KF and UKF have improved a lot over them, but still struggle with multimodal cases

Images