Current Statistics Model and Adaptive Kalman Filter in GPS Navigation
Yingchun Song · 2005
Kalman Filters are used extensively in data management of GPS dynamic positioning,while the application of Kalman Filters requires that the dynamic model(function model) and the stochastic model is reliable and practical,but in actual measure positioning it is difficult to guarantee the regular movement of the object under observation,thus easily occur the model errors.In view of such a problem in GPS dynamic positioning,this paper has discussed the Kalman Filter in practical applications that has model errors,and introduced a self-adapt Kalman Filtering computation method based on covariance matching technique,by which R can be estimated accurately when Q is known.What makes the method distinctive is that it is easy to understand the principle,and also very easy to realize in practice.It calculates the estimated value of the noise statistic through the evidence of verifying divergence,and then computes the covariance matrix of the a posterior sequence based on the estimated value of the noise statistic,thus eliminates the filtering divergence phenomena.