2024-11-222024-11-222024Chillitupa-Palomino, L., Barrera-Vargas, C.A., García-Palacios, J.H., Muñoz Díaz, I. y Naranjo Pérez, J. (2024). Machine learning-based clustering and regression for assessing temperature-induced variations on external tendon’s natural frequencies. https://doi.org/10.1201/9781003483755-274.978-103277040-6https://hdl.handle.net/11441/164800© 2024 The Author(s)Vibration-based techniques for external post-tensioning tendons have been used as non-destructive technique to assess their mechanical behavior. Thus, the continuous monitoring allows to assess their deterioration process through the continuous tracking of their natural frequencies, which may be used as damage indicators and for continuous force estimation. Hence, temperature dependencies on the frequency spectra of natural frequencies may mask potential damage among other problems such as double peaks or the existence of bracing systems. In this sense, this paper presents a mixed-cascade clustering for the frequency tracking together with the application of regression models (including linear regression and nonlinear autoregressive with exogeneous input neural network) to the classified data for identification of temperature dependencies. The proposed methodology has been applied to a case of study corresponding to several monitored tendons on a railway continuous concrete bridge.application/pdf8 p.engAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/conferenceObjectinfo:eu-repo/semantics/openAccess10.1201/9781003483755-274