A Unidirectional Trend IMM Method for SINS/DVL/USBL Navigation System Under the Time-Varying Noise Environment

In the complex underwater environment, the strapdown inertial navigation system (SINS)/Doppler velocity log (DVL)/ultrashort baselines (USBL) integrated navigation system is susceptible to the time-varying nature of measurement noise, thereby leading to a degradation in navigation accuracy. To address this issue, this article presents a unidirectional trend interacting multiple model (TIMM) method, which employs an adaptive model set adjustment strategy based on dynamic sensitivity, enabling the rapid estimation and real-time tracking of the statistical characteristics of measurement noise. Moreover, to resolve the problem of model switching lag, the Markov transition probability matrix (TPM) and the model estimation probability are dynamically adjusted using the trend of posterior information, and it integrates the maximum correntropy Kalman filter (MCKF) and the interacting multiple model (IMM). The proposed algorithm overcomes the structural constraints of traditional IMM algorithms. Through simulation experiments and river tests, the results demonstrate that compared with the extended Kalman filter (EKF), IMM, and extended IMM (EIMM), the horizontal position accuracy of the proposed method is enhanced by 14.31%, 16.38%, and 17.19%. This effectively improves the adaptability of the navigation system to harsh and unknown environments and real-time positioning accuracy.