“…Considering that failure of engineering structures mainly arises from the expansion of micro or macrocracks inherent in the applied materials under operational environments, tracing the whole evolution of all cracks and instantaneously evaluating their significance would be the essential requirement for SHMS. 45,46 With this regard, the use of the ML technology could be one of the possible substitutive and substantial methods for SHMS due to its relative simplicity but effectiveness in detecting arrested or propagating crack tips and assessing the in situ structural state by means of fracture parameters, such as SIFs determined from the cumulative ML fringe patterns in front of static or dynamic cracks. Given the additional prominent features of the ML technology, such as being a whole-field reaction, having a short response time, representing a noncontacting measurement, and exhibiting cost effectiveness, etc., this newly suggested theoretical and empirical method for the evaluation of conventional fracture could be considered as one of breakthroughs in the research field of ML-based structural diagnosis.…”