Abstract:Aiming at the problems of highly nonlinear functional functions and long time consumption in finite element calculation in the reliability analysis of rigid-frame bridges during construction period, a reliability analysis method based on the WOA?SVM is proposed. Key random variables are taken as the inputs, and the functional function value of the critical section at the maximum cantilever state is taken as the output. The WOA is employed to adaptively and globally optimize the penalty factor, insensitive coefficient and kernel function parameter of the SVM, thereby establishing a WOA?SVM surrogate model, which is then combined with the Monte Carlo method for reliability calculation and sensitivity analysis. An engineering example demonstrates that the prediction accuracy of the optimized model is significantly superior to the default parameter combination. The reliability index obtained by WOA?SVM is 3.98, satisfying the code requirement, and the relative error in failure probability is only 4.8% compared with the finite element results. Concrete tensile strength is identified as the primary factor affecting construction?stage reliability, followed by concrete density and construction load. The proposed method can provide a reference for the construction safety assessment of similar bridges.
