Unsupervised Data-driven Tool for Track Support Conditions Assessment – First Results

Résumé fourni par la source

Taking advantage of recent developments in machine learning procedures, railway researchers have been creating new tools and methodologies to improve upon structural health monitoring solutions that were developed over the last century.These new tools are now being used to assess a wide range of railway related scenarios.However, current vehicle-based railway monitoring methodologies cannot provide reliable data on track subgrade conditions.This is a key issue since subgrade condition significantly influences track dynamic response and overall track support conditions.An alternative and novel methodology to assess railway track support conditions is now under development and validation, which is based on modal analysis of the characteristic frequencies of the multi-element system composed by an instrumented railway vehicle and the railway infrastructure under assessment.Furthermore, an unsupervised data-driven procedure is currently being developed to enhance the capabilities of the proposed track monitoring methodology to automatically extract adequate results from collected data.This tool is also expected to improve the overall reliability of the proposed methodology to work with more complex data sets.To ensure the intended continuous assessment of the railway track, the developed tool is formed by a sequential combination of four steps, applied in a sliding window process over the collected input data, to reach the intended results.The four steps are, in order of application, a feature extraction step, a feature modelling step, data fusion and a feature discrimination step.This paper provides an overall description of this tool and on the obtained preliminary results, which are based on numeric simulations performed using the Simpack® software.