On the Development of Digital Twins for Condition Monitoring of Mechanical Systems

Abstract

The implementation of a condition-based maintenance policy for railway locomotives requires the development of models and methods to monitor the condition of bogie components using computer simulations, focusing on bogie frame damage and degradation of elements of the suspension system. First, the computational models are verified through the comparison of their results against the actual vehicle response, recorded in an experimental measurement campaign. The variability of the operating conditions and the parameter uncertainty are considered in the definition of the simulations using experimental design techniques. These strategies also contribute to solve the problem of the high computational cost associated with simulations of highly non-linear railway dynamics using detailed vehicle models. The post-processing involves a variety of methods including the analysis of signals in the frequency domain using the concept of transmissibility, in addition to the use of regression and statistical tools to model the vehicle response in the time domain. Condition classification methodologies are proposed to detect and locate damage on the bogie frame and the failure of suspension elements. The final result is the successful development of digital twins for the condition monitoring of the locomotive bogie structural heath and for the suspensions. In the process a methodological framework is presented to allow the use of the developments obtained in other applications for which the identification of physically based digital twins are of importance