This paper describes a non-invasive inspection technique for the estimation of longitudinal stress in continuous welded rails (CWR) to infer the rail neutral temperature (RNT), i.e., the temperature at which the net longitudinal force in the rail is zero. The technique is based on the use of finite element method (FEM), vibration measurements, and machine learning (ML). FEM is used to model the relationship between the boundary conditions and the longitudinal stress of any given CWR to the vibration characteristics (mode shapes and frequencies) of the rail. The results of the numerical analysis are used to train a ML algorithm that is then tested using field data obtained by an array of accelerometers polled on the track of interest. In the study presented in this article, the proposed technique was proven in the field during an experimental campaign conducted in Colorado. A commercial FEM software was used to model the rail track as a short rail segment repeated indefinitely and under varying boundary conditions and stress. Three datasets were prepared and fed to ML models developed using hyperparameter search optimization techniques and k-fold cross validation to infer the stress or the RNT. The frequencies of vibration were extracted from the time waveforms obtained from two accelerometers temporarily attached to the rail. The results of the experiments demonstrated that the success of the technique is dependent on the accuracy of the model and the ability to properly identify the modeshapes. The results also proved that the ML was also able to predict successfully the neutral temperature of the tested rail by using only a limited number of experimental data for the training.
9 March 2023 in Transportation Research Record: Journal of the Transportation Research Board
An inspection technique for the estimation of longitudinal stress and rail neutral temperature (RNT) in continuous welded rails (CWR) is presented. The technique is based on the use of finite element analysis (FEA), vibration measurements, and machine learning (ML). The FEA is used to establish the dependency of the vibration characteristics (mode shapes and frequencies) of any given CWR to the conditions of the track and the longitudinal stress of the rail. The dependency is then used to train an ML algorithm that is tested using experimental data obtained with two accelerometers bonded to the rail of interest, during a field test conducted in Colorado, U.S. A commercial finite element (FE) software was used to model the rail track as a short segment repeated indefinitely and under varying boundary conditions and stress. Three ML models were developed using hyperparameter search optimization techniques and k-fold cross-validation to infer the stress or RNT from the frequencies of vibration extracted from the time waveforms recorded by the accelerometers. The results demonstrated that the success of the technique is highly dependent on the accuracy of the FE model. The results also demonstrated that ML was mostly able to learn from the experimental data and successfully predict the neutral temperature.
29 November 2024 in Research in Nondestructive Evaluation