Research on the Combined Positioning Method of Maglev Trains Based on the Improved Adaptive IMM Algorithm

This paper proposes an integrated positioning method for maglev trains based on an improved adaptive interactive multi-model (IMM) algorithm to enhance the positioning accuracy of maglev trains. Firstly, according to the operating environment of maglev trains, a combined positioning scheme comprising the Global Navigation Satellite System (GNSS), Laser Doppler Velocimetry (LDV), and Inertial Measurement Unit (IMU) is designed for maglev trains. Secondly, to address the issue of measurement anomalies, a detection factor is constructed through Chi-square test (CST), enabling adaptive measurement updates. Concerning the fixed model transition probability problem, the difference in probabilities between sub-models is utilized to adaptively correct the transition probability matrix (TPM). Additionally, a sliding window is introduced to monitor changes in sub-model probabilities, further refining the transition probabilities of matching models. Finally, the proposed algorithm is compared with other IMM algorithms, achieving 15.9% and 10.1% improvements in positioning accuracy, respectively. During actual motion state changes of the maglev train, the algorithm is capable of rapidly and accurately switching to the matching model.