Scholarlink

<p>This paper tackles the challenge of numerous false targets in multi-target bearing-only passive localization under featureless conditions by proposing a maximum likelihood method enhanced with intelligent multiple motion model recognition. Unlike conventional approaches reliant on a fixed motion assumption, the proposed method utilizes a Transformer-LSTM hybrid network to evaluate target position estimates and sonar bearing measurement likelihood matrices across different motion models, and subsequently outputs localization results under mixed motion conditions. This adaptive discrimination mechanism enables the suppression of false target interference more effectively than traditional maximum likelihood or single-model strategies. Simulation experiments and real sea trial data demonstrated that the proposed method achieved an association accuracy of over 80% for bearing measurements from two passive sonars and significantly reduced localization errors. These results confirm that the proposed approach provides a practical and robust solution for multi-target passive localization in complex underwater environments, with strong potential for engineering applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Maximum Likelihood Passive Localization Method Based on Multiple Motion Model Intelligent Recognition

  • 期刊论文

  • 发表日期：

    2025年10月16日

查看全文
  • Xueli Sheng,
  • Yan Wang,
  • Bingyu Shi,
  • Linna Wan

摘要

This paper tackles the challenge of numerous false targets in multi-target bearing-only passive localization under featureless conditions by proposing a maximum likelihood method enhanced with intelligent multiple motion model recognition. Unlike conventional approaches reliant on a fixed motion assumption, the proposed method utilizes a Transformer-LSTM hybrid network to evaluate target position estimates and sonar bearing measurement likelihood matrices across different motion models, and subsequently outputs localization results under mixed motion conditions. This adaptive discrimination mechanism enables the suppression of false target interference more effectively than traditional maximum likelihood or single-model strategies. Simulation experiments and real sea trial data demonstrated that the proposed method achieved an association accuracy of over 80% for bearing measurements from two passive sonars and significantly reduced localization errors. These results confirm that the proposed approach provides a practical and robust solution for multi-target passive localization in complex underwater environments, with strong potential for engineering applications.