An Adaptive RBF Hybrid Kalman Filter for Steer-by-Wire Vehicle State Estimation With Sensors Communication Delay

The accuracy of the state sensor signals has a great impact on the tracking performance of steer-by-wire (SBW) vehicle. However, communication network load variations can cause sensor noise to undergo uncertain time-varying mean and variance, although it conforms to a Gaussian distribution. To solve this problem, this paper proposes an adaptive RBF hybrid Kalman filter. The influence of communication network occupancy on the statistical characteristics of sensor measurement noise is analyzed. In this algorithm, the Schmidt decomposition is employed, which eliminate the impact of measurement deviation caused by communication delay on the estimation accuracy. Sage-Husa noise estimator corrects the noise covariance based on the innovation sequence in the iterative process to improve the accuracy of the algorithm in the complex communication network environment. The adaptive smoothing factor radial basis function neural network adjusts the weights of the Sage-Husa noise estimator according to the occupancy and state change rates of the communication network to improve the algorithm’s responsiveness to time-varying communication network loads. Hardware-in-the-loop test results show that the proposed method effectively improves the state estimation accuracy of the SBW vehicle in a complex communication network environment.