Hybrid Rail Track Quality Analysis using Nonlinear Dimension Reduction Technique with Machine Learning

Track Geometry parameters from rail track inspection are regulated within unique safety limits for different classes. This paper focuses on developing an index that combines and quality because of the inefficiency having corrective maintenance activities between routine cycles when federal geometry violated. combination is achievable by summarizing multivariate parameters, as improvement to previous linear approaches address problem inefficient programs. The use nonlinear dimension reduction (T-Stochastic Neighbor Embedding-T-SNE) Hybrid Quality Index development, influence time-based evaluated in this study. Results show probability defects correlated with principal components but T-SNE had best prediction train-test splits despite its poor performance a blind validation set. absence observable correlation acceleration data calls further investigation.