.png){kind=icon}
.png){kind=icon}
Return
Adaptive UKF Algorithm Based on LSTM and Credibility
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ifacol.2025.11.109.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To solve the problem of mismatch between the system model and the actual model in the state estimation process, this paper proposes an adaptive UKF algorithm based on LSTM and credibility. First, the LSTM neural network is used to model the nonlinear transfer function dynamically to solve the problem of inaccuracy of nonlinear transfer function. Second, the measurement error covariance of the UKF algorithm is analyzed in the framework of credibility, and a correction method for the measurement error covariance is proposed in combination with credibility to solve the problem of mismatch of noise covariance in the filtering process. Finally, the UKF algorithm is combined with the nonlinear transfer function and the adjusted set of measurement error covariance to estimate the target state and evaluate the credibility. The results of simulation experiments show that the algorithm can effectively improve the filtering prediction and tracking performance. Copyright (c) 2025 The Authors.
Keywords:
Kalman filtering
Credibility
UKF
LSTM
State estimation