Research on an Improved Adaptive IMM-IAEKF Algorithm for Train Speed Measurement and Positioning

To address the insufficient positioning accuracy of trains, this paper adopts a technical architecture combining Beidou navigation, gyroscopes, wheel speed sensors, and radar. Aiming at the problems of model switching lag and low accuracy in the standard Interacting Multiple Model (IMM) algorithm, an improved adaptive IMM algorithm is proposed. Firstly, to tackle the model switching lag caused by the fixed transition probability matrix in the standard IMM, a correction function based on a polynomial function and a feedback structure (a first-order IIR filter) are proposed to adaptively adjust the transition probabilities. This enhances the probability of the matching model and suppresses noise effects. Secondly, to address the performance degradation of the Kalman filter due to the instability of Beidou measurement noise, an adaptive Kalman filtering algorithm based on innovation estimation is proposed. This algorithm estimates the innovation variance using maximum likelihood estimation and directly incorporates it into the gain calculation. Simulations demonstrate that the proposed algorithm outperforms other adaptive IMM algorithms in terms of transition probability robustness and state estimation accuracy.