Prior Fusion and Update Methods and Prior Complementary Methods for Two-Layer and Multi-Layer Structures

Abstract

A prior fusion and update method of a two-layer structure, including: setting the initial value of the sample number to S(0) and the sample increase step size to b; obtaining the updated system prior P(0); setting the sample number S(1 ), and make S(1)=S(0)+b; obtain the updated system prior P(1); calculate the symmetric relative entropy D(1) between the system prior P(1) and P(0); Define the number of cycles as i, and let i=1; set the number of samples S(i+1), and let S(i+1)=S(i)+b; obtain the updated system prior P(i+1) ;Calculate the symmetric relative entropy D(i+1) between the system prior P(i+1) and P(i); check whether the values ​​of symmetric relative entropy D(i+1) and D(i) are not greater than Set value e, if yes, output updated system prior P(i+1); otherwise, set i=i+1, return to step 7. The invention also discloses a prior fusion and update method of multi-layer structure, and also discloses a prior supplement method.

Description

technical field [0001] The invention relates to the technical field of satellite system modeling, in particular to a priori fusion and update method and a priori supplementary method for two-layer and multi-layer structures. Background technique [0002] When modeling the uncertainty of a satellite system, there are often a variety of prior information in the system. These prior information are mainly empirical data, previous experimental data and theoretical analysis results of the system; and how to carry out these prior information Reasonable fusion has an important impact on the uncertainty modeling of satellite systems. [0003] The existing prior information fusion methods are mainly Bayesian Melding Method (BMM) and Iterative updating procedure based on Bayesian Melding Method (I-BMM). When the fusion method models the uncertainty of the system, it first uses the structural relationship of the system to fuse the prior information of the system layer and the subsystem...

AI Technical Summary

Problems solved by technology

[0005] The iterative prior fusion and update method cannot change the updated prior distribution, and in the process of uncertainty propagation, errors will continue to accumulate, resulting in low accuracy of the updated prior distribution; at the same time, due to the I-BMM and BMM methods Both involve sampling, and in practical applications, it is difficult to determine the appropriate number of samples to obtain a balance between accuracy and efficiency; in addition, BMM and I-BMM are mainly aimed at two-tier structure systems, and cannot be used for multi-level prioritization of complex satellite systems. experience integration and update

Images