病态场景下多传感器系统误差的岭估计方法

多传感器系统误差估计是数据融合系统获得性能优势的关键前提之一。针对病态场景下传统系统误差估计方法数值不稳定的问题,对目标密集型和传感器密集型两种典型病态场景进行了理论分析,提出了多传感器系统误差的岭估计方法,以牺牲估计器无偏性的代价来改善估计结果的数值稳定性。通过引入条件数约束,给出了岭参数的最优取值方法。仿真结果表明,所提岭估计器在良态场景下与传统最小二乘估计器性能保持一致;在目标密集型场景下,与传统方法相比有显著性能优势;在传感器密集型场景下,对距离系统误差的估计性能有明显改善。

Multi-sensor systematic bias estimation method in ill-conditioned scenarios on the basis of ridge estimation#br#

Multisensor bias estimation is a key precondition for the data fusion system to achieve performance superiority. Tratitional systematic bias estimation methods are numerically instable when applied into the ill conditioned scenarios. Theoretical analysis is carried out on ill conditioning for two representative ill conditioned scenarios, i.e., the tense target scenario and the tense sensor scenario. Then the systematic bias estimation method is proposed based on ridge estimation, which improves the numerical stability of the estimation results by relaxing the constraint of estimation unbiasedness. The approach of selecting the optimal ridge parameter is given under the constraint of the condition number. Simulation results demostrate that the propsed method is consistent with the lesat squares under good conditioned scenarios, while it is superior to the traditional methods under tense target scenarios. In the case of tense sensor scenarios, the proposed method shows better performance on the range bias estimation.