基于深度加权的稳健Kalman滤波方法

在分析粗差对Kalman滤波器性能影响的基础上,通过将滤波新息的加权方式改进为深度加权平均,提出了一种基于Kalman框架的新型的稳健滤波算法.该算法仅需引入一个样本深度及权函数的计算步骤,无需针对测元的粗差检择,直接调节各测元对滤波状态的贡献.深度加权滤波扩展了传统Kalman滤波的最小均方误差优化准则,充分利用了不同测元间的相关性和测元与状态的相关性,可以有效降低含粗差数据对滤波结果的影响程度.在稳健性分析的基础上,数值算例验证了算法的可行性和有效性.

Research on robust Kalman filter based on depth-weighted

After the analysis of cumulative effect on filter results of gross errors,a new robust filter under the Kalman framework is proposed by improving the weighted mode of the innovation with the depth-weighted algorithm.For the introduction of the calculation of data depth and weighted coefficients,the filter can straightforwardly adjust the contribution of the observations to the filter states without any gross error detections.The depth-weighted step can be viewed as an extension of the optimal criterion(the minimum mean square error,MMSE) in the Kalman filter,By utilizing of the relativity of different observations as well as the relativity between the observations and the states,the new filter can effectively release the disadvantage effect on the filter results of gross errors.Based on the robustness analysis,the feasibility and the efficiency of the new filter are validated by numerical examples finally.