Railway track buckling poses a significant challenge in railway engineering, necessitating a rapid and reliable method for evaluating buckling risks to aid in its prediction and prevention. This study introduces an innovative surrogate model using a multilayer perceptron algorithm to streamline this evaluation process. The model offers an efficient alternative to time-consuming and computationally demanding three-dimensional track simulations while effectively incorporating the complexities of track dynamics data. The primary objective of this study is to reduce the evaluation process time while maintaining high accuracy of predictions. The methodology involves the generation of comprehensive track dynamics data derived from 3D track multibody dynamics model simulations and training the multilayer perceptron algorithm on this data. The dynamics model is a multibody system that includes a mixture of rigid bodies, flexible bodies, and nonlinear friction. Results indicate that the proposed surrogate model reduces the evaluation time by ∼98% while maintaining similar prediction accuracy, achieving 99.44% accuracy in replicating buckling scenarios identified by the 3D model. This demonstrates a significant improvement in computational efficiency without compromising prediction reliability. The study concludes that the developed model is a viable alternative tool for faster evaluation of buckling risks, laying the groundwork for advancing toward real-time evaluation of buckling risks.
Read full abstract