Machine Learning-Based Parametric Analysis of Railway Systems

Effective and meticulous management of railway infrastructure is essential to prevent accidents and minimize operation and maintenance costs.This requires comprehensive knowledge of the assets, their interactions, and the impact of each track parameter on the overall performance of the infrastructure.This study conducts extensive analyses using a previously calibrated finite element model of slab track, varying key track parameters within their typical ranges.The resulting data is then used to train and validate predictive models employing machine learning algorithms.This approach provides deeper insights and improves the prediction of track behavior, which involves numerous variables such as soil/subgrade, supporting layers, sleepers, pads, and rails.Additionally, the study considers the impact of train axle loads and service speeds, which are crucial factors affecting track performance.The findings highlight that the most influential parameters on railway infrastructure are soil properties, rail pad characteristics, and axle loads.This research can facilitate the implementation of predictive maintenance strategies for railway tracks and the development of innovative technological solutions, addressing industrial needs for cost reduction and enhancing the competitiveness of railway transport.