Spacecraft system anomaly detection method based on high-dimensional space mapping

Abstract

The invention discloses a spacecraft system anomaly detection method based on high-dimensional space mapping, and belongs to the technical field of space. According to the invention, a system-level fault correlation model of mutual influence among multiple telemetry parameters is established. According to the method, the cosine similarity or the Pearson correlation coefficient is used for calculating the similarity between different telemetry parameters, a clustering algorithm based on a similarity matrix is used for clustering all the telemetry parameters, all the telemetry parameters can bedivided into different subsets, and a series of subsystems are formed. A principal component analysis method is used for processing a multi-dimensional space of a subsystem, the multi-dimensional space is mapped to a new space, the dimension of the new space must be lower than the spatial dimension of original data, anomaly detection is carried out on data of each dimension in the new space, and therefore a set of suspected anomaly data points under each feature dimension can be obtained. The suspected abnormal points in all the new spatial data are integrated to obtain system-level anomaly detection information.

Description

technical field [0001] The invention relates to a system-level anomaly detection method of a spacecraft based on high-dimensional space mapping. Aiming at the multi-parameter correlation spacecraft system-level anomaly problem, this method proposes a spacecraft system-level anomaly detection method based on high-dimensional spatial mapping, and establishes a system-level correlation fault model that interacts with multiple telemetry parameters. Background technique [0002] As a typical complex system, the spacecraft system is also a high-risk field. Even a small failure of a local link may cause huge losses or disasters. [0003] In order to provide decision-makers with higher-level system-level correlated fault anomaly information, after the anomaly analysis of a single telemetry parameter is completed, it is necessary to establish a correlation model that can analyze system-level correlated anomalies caused by the mutual influence of multiple telemetry parameters. Theref...

AI Technical Summary

Problems solved by technology

If only the abnormality of a single parameter is analyzed and processed, once the spacecraft system fails, multiple telemetry parameters may be abnormal at the same time. Although these individual abnormal information can help ground managers to identify faults, they also give ground management People cause a lot of pressure, which affects their ability to make correct decisions

Images