Data driven model predictive control (DMPC) has become an effective trajectory tracking approach for unmanned surface vessel (USV). However, its offline-trained prediction models and event-triggered (ET) mechanisms result in reduced control accuracy and efficiency under time-varying ocean disturbances. To effectively improve the control accuracy and efficiency of the DMPC method under environmental influences, a two-step ET DMPC (TETDMPC) algorithm is proposed. Traditional mechanism prediction models fail to effectively account for time-varying USV dynamics, which can degrade the accuracy of control algorithms. To enhance the control accuracy of the algorithm, a v-support vector regression (vSVR)-based data driven model is introduced into the model predictive control algorithm to replace mechanism prediction model. However, the rolling horizon calculations of the DMPC and the online updates of the vSVR suffer from extensive computation. To improve the control efficiency and enable vSVR model online updates, a two-step ET mechanism is designed. The performance of the online vSVR model depends on the quality of the training data. Therefore, a spatiotemporal data updating algorithm is also developed to construct the model training dataset online. The trajectory tracking performance of the proposed TETDMPC algorithm is validated based on the CyberShip II and Dolphin 1 USVs, respectively. The results demonstrate that the TETDMPC algorithm has desirable control performance.

Multi-fidelity modeling is a common approach for predicting the drag coefficient of airfoil, offering high accuracy and low computational cost. Low-fidelity data are acquired by simplifying high-fidelity models. However, existing multi-fidelity modeling techniques fail to consider the relative importance of different low-fidelity samples and treat them uniformly, which reduces the accuracy of multi-fidelity models. To effectively utilize different low-fidelity samples, this paper proposes a hierarchical clustering weighted Gaussian process regression (HCWGPR) low-fidelity model and applies it to the multi-fidelity modeling of the airfoil drag coefficient. First, principal component analysis is used to extract the key information of low-fidelity samples. Then, empirical mode decomposition is employed to decompose the key information into high-frequency energy fluctuations and low-frequency feature vectors. Next, a hierarchical clustering weight calculation algorithm based on high-frequency energy differences and low-frequency feature similarity is proposed to assign weights to low-fidelity samples. Finally, a weighted Gaussian process regression framework is used to construct the HCWGPR. A Gaussian process regression is also adopted as a discrepancy model. The effectiveness of the proposed method is verified by 9 sets of analytical test cases and the drag coefficient prediction of 6 airfoils.

Ship dynamic models serve as the foundation for designing ship controllers, trajectory planning, and obstacle avoidance. Support vector regression (SVR) is a commonly used nonparametric modelling method for ship dynamics. Achieving high accuracy SVR models requires a substantial amount of training samples. Additionally, as the number of training samples increases, the computational efficiency for solving the quadratic programming problem (QPP) of SVR decreases. Ship controllers demand dynamic models with both high accuracy and computational efficiency. Therefore, to enhance the prediction accuracy and computational efficiency of SVR, this paper proposes a nonparametric modelling method based on twin SVR (TSVR). TSVR replaces a large QPP with a set of smaller QPPs, significantly enhancing generalizability and computational efficiency. To further improve the predictive accuracy of TSVR, the puma optimizer algorithm is employed to determine the optimal hyperparameters. The performance of the proposed method is validated using a Mariner class vessel. Gaussian white noise is introduced into the modelling data to simulate measurement error. The TSVR model accurately predicts various zigzag and turning circle manoeuvring motions under disturbance conditions, demonstrating its robustness and generalizability. Compared to the SVR model, the TSVR model achieves lower root mean square error and computational time, confirming its superior predictive accuracy and computational efficiency.

Traditional Artificial Potential Field (TAPF) method can only work out a single solution when solving path planning problem. This paper proposes a method to obtain multiple solutions by reconstructing the potential field by means of a new repulsive function. Firstly, the attractive function is simplified to improve the unbalanced attraction in traditional potential field. Next, a line-obstacle model is introduced to modify the structure of the repulsive potential field. Finally, the path planning with multiple solutions can be achieved through a multi-round iterative strategy. After MATLAB simulations conducted, the results demonstrate that this strategy can achieve high success rates for multi-path planning in a single scenario, with smooth paths and strong robustness.