基于机器学习的多算法融合航迹稳健起始方法

A Robust Multi-Algorithm Fusion Track Initiation Algorithm Based on Machine Learning

针对在强电子对抗和复杂雷达任务环境中杂波、干扰等影响目标航迹正确有效起始的问题,提出了一种多算法融合学习航迹稳健起始方法。该方法将航迹起始问题视为分类问题,使用经典的机器学习分类算法——随机森林和梯度提升决策树(gradient boosting decision tree,GBDT)为基础进行融合分类。考虑将两种方法进行Chair-Varshney最优决策融合,实现对目标航迹的高效正确起始。通过仿真实验将本文提出的方法和随机森林、GBDT、启发式规则等经典方法进行对比,结果表明:多算法融合学习航迹稳健起始方法的整体性能更好,显著优于启发式规则航迹起始方法和GBDT航迹起始方法。

A robust track initiation algorithm based on multi-algorithm fusion learning is proposed to resolve the correct and effective track initiation issue due to the effect of clutter and jamming in strong ECM and complex radar mission environment.This method regards the track initiation issue as classification issue,and uses the classical machine learning classification algorithm-random forest and GBDT as basis for fusion and classification.Chair-Varshney optimal decision fusion is applied to these two methods to achieve the effective and correct initiation of the target track.The simulation is used to compare the proposed method in this paper with random forest,GBDT,and Heuristic rule.The results show that the robust track initiation algorithm based on multi-algorithm fusion learning has better overall performance,much better than that based on GBDT and Heuristic rule.