An Innovation-Based Adaptive Cubature Kalman Filtering for GPS/SINS Integrated Navigation
Chengying Song, Zhuo Huang, Yifei Wu, Sheng Li, Qingwei Chen
Abstract
The adaptive filtering has always been a research focus for the inaccurate model, time-varying noises, and abnormal measurements in practice. The major challenges faced by all adaptive filters are the selection of test criteria and design of optimal adaptive factors. Accordingly, this article presents an innovation-based adaptive cubature Kalman filtering method for satellite and inertia integrated navigation systems. Aiming at preventing outliers from contaminating the filtering process, the chi-square test first detects faults in the innovations. Also, a three-segment (T-s) compression function is utilized to degrade or isolate the abnormal innovations. In the second test, a multiple adaptive weighting matrix (MAWM) with fading memory is utilized to modify the predicted state covariance matrix (PSCM) according to the variance matching principle. Moreover, a variable threshold is designed to limit the parameter fluctuations for safeguarding the filtering stability. Through a series of Monte Carlo simulations and a practical flight experiment, the superiority of the proposed method is verified in terms of higher accuracy, stronger robustness, and better adaptability.