Robust Resilient Polarization Concurrent Orientation Model in Harsh Environments

The global navigation satellite system (GNSS) plays a crucial role in practical applications of the industrial informatics field such as patrol inspection systems and security monitoring. Nevertheless, it performs inadequately in acquiring real-time heading information. Accordingly, a robust resilient fully autonomous polarization concurrent orientation strategy is presented in this study to offer stable heading information in harsh environments. Unlike previous methods, we attempt to associate multiscale dynamic kernel principal component analysis (MDKPCA) with the concurrent orientation model to achieve real-time heading information. Specifically, MDKPCA and a kernel density estimation strategy are introduced to monitor and distinguish the raw dynamic heading data output from a polarization compass in real time. A promising hierarchical resilient anti-Hebbian long short-term memory neural network with additional time gates is incorporated to compensate for the heading error in the low-frequency data state. A resilient time-frequency peak filtering method with the advantage of a long-short window is employed to remove noise in the mixed data state composed of middle-frequency heading data and high-frequency noise synchronously. Various representative experiment results demonstrate that the proposed model is competent to remarkably ameliorate the GNSS navigation accuracy and robustness by cutting down the error affected by vehicle attitude changes and different types of noises in harsh environments.