高级检索

深海分布式多垂直阵决策级融合探测性能评估

Assessment of the fusion detection capability of the distributed vertical line arrays on decision-level in deep-ocean

  • 摘要: 随着水下目标隐蔽性和机动性的不断提高,仅靠单部被动声呐难以实现有效探测,而分布式融合技术提供了一种策略。针对深海目标探测场景,文章基于被动声呐方程和能量检测法,将“与”(AND)、“或”(OR)、多数投票(K out of N, K/N)和Chair-Varshney(CV)这4类决策级融合策略应用至分布式多节点垂直阵并对其融合探测性能进行了系统的分析与评估。结果表明,AND融合能有效降低系统虚警率,但检测率随节点数增多而显著下降;OR融合能最大程度提升系统检测率,但其虚警性能过于依赖单节点虚警率;K/N融合系统的性能介于AND和OR融合之间,但其检测性能会随各节点距离增大而急剧下降,适用于节点数较多的系统;CV融合作为理论最优融合策略,其检测率和虚警率仅略低于OR融合和ANR融合,且相较K/N融合对远距离目标具有更高的检测概率,但其局限在于需要获得各节点性能参数的先验信息。

     

    Abstract: In recent years, it has become increasingly difficult for individual passive sonars to detect targets due to the continuous improvement in the stealthiness and mobility of submerged targets. However, distributed multi-sensor fusion technology offers a promising solution. In this study, four decision-level fusion strategies (AND, OR, K/N, and CV) are applied to distributed multiple vertical line arrays for target detection scenarios in deep-ocean environments. Their fusion detection capabilities are analyzed and evaluated based on the passive sonar equation and the energy detection method. The results indicate that AND fusion can effectively reduce the false alarm rate (FAR), but it is accompanied by a significant decrease in detection rate (DR) as the number of vertical line arrays increases. OR fusion can greatly improve the DR, although its false alarm performance heavily depends on that of individual vertical line arrays. The K/N fusion system performs between the above two systems; however, its DR decreases sharply with increasing spacing between vertical line arrays, making it suitable for systems with densely distributed vertical line arrays. As a theoretically optimal fusion strategy, CV fusion achieves a DR and FAR slightly lower than those of OR and AND fusion, with higher detection probability for distant targets compared to K/N fusion. However, its limitation lies in requiring knowledge of the performance parameters of each vertical line array.

     

/

返回文章
返回