高速铁路既有路基沉降代理模型搭建及行车性能预测分析OA

This paper proposed a surrogate model based on Radial Basis Function Neural Networks(RBFNN)to enable rapid calculation of high-speed train operational safety and ride smoothness under multi-operating conditions induced by subgrade settlement.By conducting a literature review and combining it with an optimized Latin hypercube sampling method,this paper constructed a uniformly distributed sample space for subgrade settlement.An automated simulation workflow was established using the integrated Abaqus-Isight-Python co-simulation approach,enabling batch input of settlement data and automatic output of train operation performance.Based on this dataset,an RBFNN-based surrogate model for high-speed railway subgrade settlement was developed to predict train operation performance.Results show that:the proposed surrogate model achieves lower relative errors compared to other commonly used surrogate models.Through sensitivity analysis,the variation patterns of output variables with respect to input variables are obtained.It is found that the settlement wavelength L predominantly influences all performance indicators of train operation,and significant interaction effects exist among the parameters.When the settlement wavelength is 26.35 m and the lateral differential settlement factor is 1,the critical threshold of settlement amplitude reaches its minimum value of 9.19 mm.