Distributed Robust Particle Filter Based on Wasserstein Metric

In this paper, for nonlinear systems with uncertain interference factors, particle filter can no longer accurately estimate the state when both process noise and measurement noise are unknown, so a distributed robust particle filter method based on Wasserstein metric is proposed. The method mainly uses Wasserstein metric to construct the fuzzy set of real data distribution, replaces the original empirical distribution with the prior estimation distribution of particle filter, and takes it as the center point of the new fuzzy set; Secondly, on the premise of tolerable conservatism, the optimal state estimator in this state is constructed according to the criterion of minimum root mean square error in the worst case; Then, the updating and resampling of particle filter are used to complete the calculation of a posteriori estimation, which solves the problem of state estimation of nonlinear systems under uncertain disturbances; Finally, an example is given to verify the feasibility and effectiveness of the distributed robust particle filter method.