GPS Navigation Using Fuzzy Neural Network Aided Adaptive Extended Kalman Filter

TLDR: The fuzzy method combined with NN to identify the noise covariance matrix is proposed and Numerical simulations show that the adaptation accuracy based on the proposed approach is substantially improved.

Abstract: GPS navigation state processing using the extended Kalman filter provides optimal solutions (in the mean square sense) if the noise statistics for the measurement and system are completely known. Covariance matching method is a conventional adaptive approach for estimation of noise covariance matrices. This innovation-based adaptive estimation shows noisy result if the window size is small. To overcome the problem, the fuzzy method combined with NN to identify the noise covariance matrix is proposed. The structure of FNN can automatically identify the fuzzy rules and tune the membership functions by modifying the connection weights of the network using back-propagation algorithm. Numerical simulations show that the adaptation accuracy based on the proposed approach is substantially improved.