Heave motion prediction with variational bayesian-based strong tracking Kalman filter

Real-time prediction of heave motion plays a crucial role in the control design of hybrid active-passive heave compensation (HAHC) systems, which compensate the time delay between the sensor and the actuator of the HAHC. Most studies encounter challenges in effectively balancing noise estimation and algorithm convergence. This study establishes a parametric predictive model by employing a nonlinear tracking differentiator (NTD) and the fast Fourier transform (FFT), which is able to capture the characteristics of the heave motion. Then a novel variational Bayesian based strong tracking Kalman filter (VBSTKF) is designed to cope with the model uncertainties, and to obtain accurate process and measurement noise matrices. To verify the accuracy and convergence of the proposed algorithm, we apply both simulation data with a sine wave and real ship experiment data. Additionally, we establish an experimental platform for real-time testing of the proposed predictor to confirm its significant impact on HAHC. All the results show that the prediction errors of the proposed predictor decrease greatly compared with previous studies, with good robustness of model uncertainties and sensor noises. Especially, the prediction error of the proposed algorithm remains below 0.02m with a 1.5m wave amplitude. Meanwhile, the proposed predictor gives an excellent heave compensation performance in real-time testing, indicating its potential for application under actual sea conditions to compensate for heave motion in HAHC systems. • A variational Bayesian-based strong tracking Kalman filter (VBSTKF) for generating predictive trajectories of heave motion. • A novel VBSTKF algorithm is proposed to identify the parameters of the predictive model. • An approximate heave predictive model is established using fast Fourier transform to capture the frequency characteristics. • The prediction error of the proposed algorithm remains below 0.02m with a 1.5m wave amplitude. • Extensive comparative experimental results validate the performance of the proposed predictor for heave compensation.