Directional and adaptive forgetting factor based recursive least square algorithms for maneuvering dynamic identification and motion prediction of unmanned surface vessel

In recent years, Unmanned Surface Vessel (USV) has been increasingly widely used in commercial and scientific research fields. The prerequisite of realizing intelligent control and auxiliary decision of USV is to establish accurate mathematical model of maneuvering motion and carry out effective parameter identification. In this paper, to overcome the limitations of traditional recursive least squares (FFRLS) in dealing with non-continuous excitation and dynamic change environment, two improved recursive least squares algorithms: directional forgetting recursive least squares algorithm (DFFRLS) and adaptive forgetting recursive least squares algorithm (AFFRLS) are proposed. In this paper, the training datasets are established by validated CFD simulated turning test and Z-shaped maneuverability test. The parameters of the model are identified by DFFRLS and AFFRLS algorithm, and compared with the traditional FFRLS algorithm. The results show that DFFRLS has higher identification accuracy and robustness through directional decomposition when dealing with abnormal data such as random noise, sensor fault and data mutation. However, the AFFRLS, by adapting the forgetting factor adaptively, makes it possible to be more accurate in the process of data updating, especially when the parameters change rapidly. The research results of this paper show that DFFRLS and AFFRLS algorithms have significant advantages in parameter identification of USVs, especially in specific complex and dynamic environments, which can effectively improve the accuracy and performance of the maneuvering model.