Mixed Kernel Regularized K-Nearest Neighbor Based Weighted Twin Support Vector Regression for Nonparametric Modeling of Ship Maneuvering Motion

In this study, an innovative method, mixed kernel regularized k-nearest neighbor based weighted twin support vector regression (MRKNNWTSVR) is designed for modeling the maneuvering motion of ships. This method effectively integrates regularized k-nearest neighbor based weighted twin support vector regression (RKNNWTSVR) with a mixed kernel function, thereby facilitating the accurate identification of a nonparametric model for ship maneuvering. The mixed kernel function, combining the attributes of the radial basis and polynomial kernel functions, significantly enhances both the accuracy and the generalization. To further increase the model's accuracy, a parameter optimization strategy based on a dung beetle optimizer has been incorporated. The model's performance and generalization potential have been validated using the KVLCC2 ship model, with data sourced from the SIMMAN 2008 database. The results from these tests demonstrate the MRKNNWTSVR model's exceptional generalization ability and its overall efficacy.