A data-driven assessment model for collision responses of offshore platform structure with ship using hybrid intelligent approaches

Abstract

The accidents of ship collision with offshore platform bring huge risks to the safety of jacket structure. This paper proposed a data-driven assessment model for collision responses of offshore platform structure by integrating several intelligent and data-driven approaches. In the model, artificial neural network combining with dynamic particle swarm optimization algorithm (DPSO-ANN) is addressed to predict the collision responses of offshore platform structure, in which the uniform design method (UDM) is adopted to select training data. Principal component analysis (PCA) is used to eliminate correlation and redundancy of multivariate data of collision responses. Further, based on grey relational analysis (GRA), a GRA-based assessment index is established for the collision responses. In the case study, the prediction results demonstrate that the DPSO-ANN model could approximate the FEA (Finite Element Analysis) well, and the results of GRA-based assessment indicate that the collision velocity and mass of ship are two key factors influencing the collision responses of offshore structure. Also, the assessment results show that it is necessary to pay more attention and vigilance to the situations of high-speed approaching of large tonnage ship and ship berthing at low water level. The proposed model could serve as a beneficial and efficient assessment tool to support safety design of offshore platform structure under collision accidents.