空间控制技术与应用2025，Vol.51Issue(4)：52-64,13.DOI:10.3969/j.issn.1674-1579.2025.04.005
基于多源信息融合的智能感知定位与威胁等级评估研究
Research on Intelligent Perception Positioning and Threat Level Evaluation Based on Multi-source Information Fusion
摘要
Abstract
The significant increase in space debris has led to a dramatic escalation in collision risks for spacecraft,necessitating the urgent development of high-precision space debris situational awareness and collision threat assessment technologies.This paper focuses on space debris targets and proposes a target perception and localization method that integrates multi-source heterogeneous observation data.By combining physical mechanism modeling with machine learning algorithms,a hybrid collision threat level assessment mechanism is constructed.At the perception and localization level,a unified mathematical model for multi-satellite observation is established,effectively integrating three types of heterogeneous observation data:distance,angle,and velocity.Subsequently,a Levenberg-Marquardt(LM)algorithm based on Huber weighting and iterative re-optimization is designed,significantly enhancing localization accuracy and algorithm robustness under abnormal data conditions.At the threat assessment level,a hybrid decision-making framework is proposed,which integrates an improved physical collision probability calculation model with a random forest classifier.This framework comprehensively considers both collision probability estimation and the impact of potential kinetic collision consequences,achieving efficient and precise classification of debris threat levels.Simulation results demonstrate that the proposed localization algorithm significantly outperforms traditional least squares estimation and geometric analysis methods in terms of root mean square error(RMSE)for position estimation.Simultaneously,the threat level classification model exhibits high overall classification accuracy,with the random forest classifier demonstrating superior discriminative performance compared to logistic regression models.This research provides an effective technical solution for space-based object surveillance missions and autonomous spacecraft collision avoidance decision support.关键词
数据融合/空间碎片/轨道确定/机器学习/分类评估Key words
data fusion/space debris/orbit determination/machine learning/classification evaluation分类
航空航天引用本文复制引用
郝雅波,白雪,徐明..基于多源信息融合的智能感知定位与威胁等级评估研究[J].空间控制技术与应用,2025,51(4):52-64,13.基金项目
国家资助博士后研究人员计划和中国博士后科学基金资助项目(BX20250465) Supported by the Postdoctoral Fellowship Program and China Postdoctoral Science Foundation under Grant Number(BX20250465) （BX20250465）
