基于稳健及有偏估计理论的多传感器误差配准方法研究

Multi-sensor bias registation, which estiamtes and compensates systematic biases of the dispersive sensors, is the key precondition for the data fusion system to achive its expected superior performance. Traditional bias registration methods, on one hand, rely on the assumption of ideal data association. Under the environment with nonideal association, the bias estimates usually depart away from their true values. On the other hand, traditional methods are sensitive to sensor-target distribution scenarios. Under the ill-conditioned scenarios, the phenomenon of serious numerical instability occurs. Focusing on the technical bottlenect problems encountered in the practical process of multi-sensor data fusion system, this project makes indepth analysis about the influence mechanism of the nonideal association and the ill-condition of scenarios on multi-sensor bias registration, studies the “special outlier” feature of the misassociations, and constructs scientific and rational evaluation index on the ill-condition of scenarios. Based on this, we draw on the experience of the basic idea and processing means of the robust estimation theory and biased estimation theory, and propose new multi-sensor bias registration approaches which can resist misassociations and ill-condition. The breakthrough of the above research content can further deepen and enrich the multi-sensor bias registration theory. The research achievements are expected to apply to the multi-active, multi-passive, and multi-active/passive sensor data fusion systems. It is of great significance to improve the robustness and adaptability of the multi-sensor data fusion system to the complicated application environments and to promote the practical process of the data fusion system.

多传感器误差配准负责对传感器系统误差进行估计与补偿，是数据融合系统获得预期性能优势的关键前提。传统误差配准方法，一方面，依赖于理想数据关联假设，非理想关联环境下的估计结果常常偏离真值；另一方面，对传感器-目标分布场景敏感，在病态场景下会出现严重的数值不稳定现象。本项目着眼于多传感器数据融合系统实用化过程中遭遇的技术瓶颈问题，深入分析错误关联与场景病态性对多传感器误差配准性能的影响机理，研究错误关联的特殊野值特征，构建科学合理的场景病态性评估指标，在此基础上，借鉴稳健估计及有偏估计理论的基本思想和处理手段，提出抗错误关联及场景病态性的多传感器误差配准新方法。上述研究内容的突破，可进一步深化和丰富多传感器误差配准理论，研究成果有望应用于多主动、多被动、多主/被动传感器融合系统中，对提升多传感器数据融合系统对复杂应用环境的稳健性及适应性,推进融合系统的实用化进程具有重要意义。

在多传感器数据融合系统中，多传感器误差配准主要是对各分立传感器存在的系统误差进行估计与补偿，该步骤是数据融合系统获得性能优势的关键环节。传统的多传感器误差配准方法通常依赖于理想的数据关联假设，而且对传感器-目标分布场景引发的病态性缺乏足够重视，在非理想关联和（或）病态场景下的误差配准性能衰退严重，进而引起融合系统的性能退化。. 本项目的主要研究内容及取得的成果包括：. 第一，在深入分析错误关联与场景病态性对多传感器误差配准性能影响机理的基础上，研究了错误关联的“特殊野值”特征，构建了科学合理的场景病态性评估指标，通过借鉴稳健估计及有偏估计理论的基本思想和处理手段，提出了一系列抗错误关联及场景病态性的多传感器误差配准新方法。. 第二，利用多目标间的结构特征信息，提出了基于参照角度序列特征的主被动传感器抗差数据关联与相对角度系统误差的稳健估计方法，有效提升了雷达/ESM数据融合系统对复杂误差环境的适应能力。. 第三，本项目提出了一种跨多精度ESM传感器数据的容差识别冲突检测方法，检测结果可作为雷达信号识别数据库优化调整的重要依据。. 通过上述研究内容的突破，本项目为多传感器误差配准提供了一些新的理论方法与技术手段，有效提升了多传感器数据融合系统的稳定工作能力以及对复杂应用环境的适应性，进一步推进了数据融合系统的实用化进程。