Year: 2020
Type: article
Abstract: Guidelines, including recommendations for sample size, are needed in the design of track inspection technology evaluation tests. This report provides and overview of recommended methods for evaluating the effectiveness of different types of track inspection systems. These methods produce performance measures that can be used to validate the technology against ground truth, which is often difficult to obtain, and to compare technologies. As sufficient test samples are needed to make valid comparisons, sample size is also addressed by, for example, grouping flaws of similar size into bins and by combining repeat inspections. The report describes each of the evaluation methods and demonstrates their application to the evaluation of a track geometry measurement system (TGMS). The relatively new Model-Assisted Probability of Detection (MAPOD) technique, which combines actual results with those from computer models of the inspection technology, could usefully be demonstrated on one or more track inspection technologies as part of future efforts.\n
Source: ROSA P
Language: English
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3045110204