Structural health assessment of high-pile wharves based on damage cause inversion

Existing methods for evaluating wharf structures are characterized by lag and inability to continuously obtain structural safety information. This study investigates damage cause inversion and integrates two common failure modes of wharves. The load redundancy coefficient is introduced as an evaluation index for wharf structures. Using machine learning methods, the study predicts the intensity, location, and load redundancy coefficient of damage causes. Results indicate that the PSO-BP (Particle Swarm Optimization Back Propagation) neural network model achieves the best performance in strength inversion and load redundancy prediction. The PSO-SVM (Particle Swarm Optimization Support Vector Machine) model excels in identifying the location of damage causes. The high predictive accuracy of the PSO-BP and PSO-SVM models in this study provides status evaluation indices for wharf health monitoring based on damage cause inversion, offering a reference for long-term and dynamic evaluation of wharf structures.