Intelligent evaluation of impact damage for subsea protective facilities with asymmetric openings

In this paper, the influence of the structural parameters of subsea protective devices with asymmetric openings is studied through intelligent damage assessments. Moreover, three back propagation (BP) neural networks, the Levenberg-Marquardt (LM) algorithm, Bayesian regularization (BR) algorithm and scaled conjugate gradient (SCG) algorithm, are selected to evaluate the damage to subsea protective facilities. Finite element models of subsea protective facilities without openings, square subsea protective facilities with openings and circular subsea protective facilities with openings are established, and finite element analysis is carried out to establish the database training algorithm. The accuracy of the database is ensured through testing to verify the finite element method. The influence of the opening, opening shape and structural damage on the damage assessment is analysed to find the most suitable intelligent damage assessment algorithm for subsea protective facilities with asymmetric openings. The results show that the opening will increase the error of damage identification, and the square opening will increase the maximum error. Meanwhile, structural damage will increase the damage evaluation error of subsea protective facilities with asymmetric openings but will reduce the damage stress evaluation error of subsea protective facilities without openings. In addition, the identification error of the BR neural network algorithm for each working condition, which is suitable for intelligent damage identification of subsea protective facilities, is very low.