Identification of Behavioural Models for Railway Turnouts Monitoring

10/15/2019

∙

This study introduces a low-complexity behavioural model to describe the dynamic response of railway turnouts due to the ballast and railpad components. The behavioural model should serve as the basis for the future development of a supervisory system for the continuous monitoring of turnouts. A fourth order linear model is proposed based on spectral analysis of measured rail vertical accelerations gathered during a receptance test and it is then identified at several sections of the turnout applying the Eigensystem Realization Algorithm. The predictviness and robustness of the behavioural models have been assessed on a large data set of train passages differing for train type, speed and loading condition. Last, the need for a novel modeling method is argued in relation to high-fidelity mechanistic models widely used in the railway engineering community.

READ FULL TEXT

Identification of Behavioural Models for Railway Turnouts Monitoring

Multi-fidelity modeling with different input domain definitions using Deep Gaussian Processes

Predicting Nanorobot Shapes via Generative Models

Multi-fidelity surrogate modeling for temperature field prediction using deep convolution neural network

Estimation of lateral track irregularity through Kalman filtering techniques

A novel reduced order model for vortex induced vibrations of long flexible cylinders

Error Approximation and Bias Correction in Dynamic Problems using a Recurrent Neural Network/Finite Element Hybrid Model

Elimination of spectral blocking by ensuring rotation-free property of discretised pressure gradient within a spectral semi-implicit semi-Lagrangian global atmospheric model

Identification of Behavioural Models for Railway Turnouts Monitoring

Related Research

Multi-fidelity modeling with different input domain definitions using Deep Gaussian Processes

Predicting Nanorobot Shapes via Generative Models

Multi-fidelity surrogate modeling for temperature field prediction using deep convolution neural network

Estimation of lateral track irregularity through Kalman filtering techniques

A novel reduced order model for vortex induced vibrations of long flexible cylinders

Error Approximation and Bias Correction in Dynamic Problems using a Recurrent Neural Network/Finite Element Hybrid Model

Elimination of spectral blocking by ensuring rotation-free property of discretised pressure gradient within a spectral semi-implicit semi-Lagrangian global atmospheric model