Abstract:In view of the issues of experience-dependent prestressing adjustment of the back tie rods in large ship lock miter gates and insufficient accuracy in predicting gate deformation under multi-condition coupling effects,the relationship between back tie rod prestress and gate deformation characteristics is investigated.An approach combined finite element analysis and machine learning is adopted.A multi-condition sample dataset is constructed through large-scale parametric finite element simulations,and a mapping model between back tie rod prestress,water level,and gate deformation is established to achieve rapid and accurate prediction of gate deformation under various operating conditions,thereby reducing repetitive finite element calculations.On the basis of the simulation results,a “bow-shaped” sensor layout scheme using weak fiber grating sensing technology is proposed,and an online monitoring system integrating real-time sensing,intelligent analysis,and visualized warning is developed.The results demonstrate that the proposed system can continuously acquire responses at critical locations of the gate and identify abnormal deformation trends,forming a technical chain of “simulation database construction-intelligent prediction-online sensing-risk warning”.The research results provide technical support for refined prestress adjustment of back tie rods,structural safety assessment,and ship lock operation and maintenance decision-making,promoting the transition of ship lock maintenance from scheduled inspection to condition-based maintenance.