Computational Technology Resources - CCC

J.A. Sainz-Aja¹, D. Ferreño¹, I.A. Carrascal¹, J. Pombo^2,3, J.A. Casado¹, S. Diego¹ and M. Cuartas⁴

¹Laboratory of Science and Engineering of Materials, University of Cantabria. E.T.S. de Ingenieros de Caminos, Canales y Puertos, Santander, Spain
²Institute of Railway Research, University of Huddersfield, United Kingdom
³IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Portugal and ISEL, IPL, Portugal
⁴Group of Information Technologies, University of Cantabria. E.T.S. de Ingenieros de Caminos, Canales y Puertos, Santander, Spain

doi:10.4203/ccc.1.20.5

J.A. Sainz-Aja, D. Ferreño, I.A. Carrascal, J. Pombo, J.A. Casado, S. Diego, M. Cuartas, "Rail pad stiffness estimation based in machine learning algorithms", in J. Pombo, (Editor), "Proceedings of the Fifth International Conference on Railway Technology: Research, Development and Maintenance", Civil-Comp Press, Edinburgh, UK, Online volume: CCC 1, Paper 20.5, 2022, doi:10.4203/ccc.1.20.5

Keywords: rail pad, machine learning, operational conditions, dynamic stiffness.

Abstract

Rail pads are interposed between the steel rails and the concrete sleepers on the railway lines to protect the latter from the impacts induced by the passage of the trains. They provide compliance to the track and play a fundamental role to maximize its durability and minimize the maintenance costs. Rail pads can be fabricated with different polymeric materials that display a non-linear mechanical behavior which strongly depends on the external conditions. Therefore, it is extremely difficult to estimate its mechanical properties, in particular its dynamic stiffness. In this work, several machine learning algorithms (multilinear regression, K nearest neighbors, regression tree, random forest, multi-layer perceptron and support vector machine) have been optimized to determine the dynamic stiffness of rail pads manufactured in EPDM, TPE or EVA, depending on the in-service conditions (temperature, frequency, axle load and toe load). A dataset consisting of 720 stiffness tests under different combinations of these variables was available for the training and testing of the models. The optimal algorithms for EPDM, TPE and EVA were, respectively, multi-layer perceptron (R2 of 0.990 and mean absolute percentage error of 6.51%), multilayer perceptron (0.994 and 2.32%) and random forest (0.968 and 4.91%).

download the full-text of this paper (PDF, 8 pages, 605 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description

	Computational & Technology Resources an online resource for computational, engineering & technology publications
	not logged in - login
Front Page Browse CCC CCP CSETS CTR IJRT Other Authors Search Purchase Guide FAQ Contact us	Civil-Comp Conferences ISSN 2753-3239 CCC: 1 PROCEEDINGS OF THE FIFTH INTERNATIONAL CONFERENCE ON RAILWAY TECHNOLOGY: RESEARCH, DEVELOPMENT AND MAINTENANCE Edited by: J. Pombo Paper 20.5 Rail pad stiffness estimation based in machine learning algorithms J.A. Sainz-Aja¹, D. Ferreño¹, I.A. Carrascal¹, J. Pombo^2,3, J.A. Casado¹, S. Diego¹ and M. Cuartas⁴ ¹Laboratory of Science and Engineering of Materials, University of Cantabria. E.T.S. de Ingenieros de Caminos, Canales y Puertos, Santander, Spain ²Institute of Railway Research, University of Huddersfield, United Kingdom ³IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Portugal and ISEL, IPL, Portugal ⁴Group of Information Technologies, University of Cantabria. E.T.S. de Ingenieros de Caminos, Canales y Puertos, Santander, Spain doi:10.4203/ccc.1.20.5 Full Bibliographic Reference for this paper J.A. Sainz-Aja, D. Ferreño, I.A. Carrascal, J. Pombo, J.A. Casado, S. Diego, M. Cuartas, "Rail pad stiffness estimation based in machine learning algorithms", in J. Pombo, (Editor), "Proceedings of the Fifth International Conference on Railway Technology: Research, Development and Maintenance", Civil-Comp Press, Edinburgh, UK, Online volume: CCC 1, Paper 20.5, 2022, doi:10.4203/ccc.1.20.5 Keywords: rail pad, machine learning, operational conditions, dynamic stiffness. Abstract Rail pads are interposed between the steel rails and the concrete sleepers on the railway lines to protect the latter from the impacts induced by the passage of the trains. They provide compliance to the track and play a fundamental role to maximize its durability and minimize the maintenance costs. Rail pads can be fabricated with different polymeric materials that display a non-linear mechanical behavior which strongly depends on the external conditions. Therefore, it is extremely difficult to estimate its mechanical properties, in particular its dynamic stiffness. In this work, several machine learning algorithms (multilinear regression, K nearest neighbors, regression tree, random forest, multi-layer perceptron and support vector machine) have been optimized to determine the dynamic stiffness of rail pads manufactured in EPDM, TPE or EVA, depending on the in-service conditions (temperature, frequency, axle load and toe load). A dataset consisting of 720 stiffness tests under different combinations of these variables was available for the training and testing of the models. The optimal algorithms for EPDM, TPE and EVA were, respectively, multi-layer perceptron (R2 of 0.990 and mean absolute percentage error of 6.51%), multilayer perceptron (0.994 and 2.32%) and random forest (0.968 and 4.91%). download the full-text of this paper (PDF, 8 pages, 605 Kb) go to the previous paper go to the next paper return to the table of contents return to the volume description
Back to top	©Civil-Comp Limited 2023 - terms & conditions