Estimation with particle filter under model uncertainty

A new method for estimation under model uncertainty is proposed. Section 1 defined the problem and introduced algorithms invented before. Section 2 pointed out the limitation of exist methods and presented ours. The core is that we treat unknown model mode as sample of an infinite model mode set. After properly modeled, section 3 presented particle filter based computation algorithm for our model. Section 4 provide a demonstration example and shows when adopt our method, model uncertainty problem can be solved with good efficiency. Section 5 concludes preliminarily that the proposed algorithm is promising.