A real-time robust filtering and fusion method based on multiple median estimation

In view of the interference of complicated electromagnetic environment and the unrobustness of traditional method in the weighted fusion of multi-source electronic measurement data, with robust statistics theory as a guide, this paper proposed a robust filter method based on real-time median estimation and constant velocity model, and applied the filter residual error scale to the dynamic fusion weights calculation under the optimal weight assignment principle, then obtained a new distributed robust weighted fusion method. With typical experimental data, the weighted fusion was tested, and the filtering method was compared with Kalman filter. The results showed that the methods proposed, which less affected by gross error, were effective and robust in practical engineering application, and achieved the adaptation to measurement noise.