Sensor fault diagnosis of GPS/INS tightly coupled navigation system based on state chi-square test and improved simplified fuzzy ARTMAP neural network

In order to improve the reliability of GPS/INS tightly coupled navigation system, a novel sensor fault diagnosis scheme combined with the advantages of state chi-square test (SCST) and simplified fuzzy ARTMAP neural network (SFAM) is presented in this paper. Firstly, in order to overcome the problem of fault detection in noisy environment, a two state propagator based on SCST is applied in the navigation system to detect single fault to multiple faults of different sensors (gyroscope, accelerometer and GPS receiver). Then, traditional grey wolf optimizer is improved by adding a memory influence term in the update formula and a 'greedy strategy' to enhance new individual's fitness. A simplified fuzzy ARTMAP neural network optimized by improved grey wolf optimizer is employed to classify sensor fault according to SCST result. Meanwhile, another improved simplified fuzzy ARTMAP neural network is employed to identify the fault magnitude and fault occurring time. Further, faults features base on SCST results corresponding to different sensors and different flight trajectories are analyzed. The analysis results are used to design fault classification and identification scheme. Finally, simulations which take into account all fault types, different fault magnitude, and different flight trajectories are given. The simulation results show that the proposed fault diagnosis method can detect, classify, and identify the sensor fault of GPS/INS navigation system.