Acoustic emission source characterisation using evolutionary optimisation

When a crack initiates and grows in a metal or composite structure, for example, due to high cycle fatigue, the crack propagation gives rise to acoustic emissions (AE)-ultrasonic waves travelling through the structure. Because the presence and rate of growth of any cracks are important pieces of information about the condition or health of the structure, the monitoring of AE activity using sensors mounted on its surface is a potentially useful technique of structural health monitoring. In tests, acoustic emissions are often simulated by breaking a pencil lead against the surface of the structure in a standardised way (a "Hsu-Nielsen" source), but the forces that this imparts are not well understood at present. The current paper proposes a new evolutionary optimisation-based approach to source characterisation. The principle is to introduce a parametrised representation of a general source and then identify the parameters that allow the source to best match responses measured elsewhere on the structure. The predicted responses are modelled using a local interaction simulation approach (LISA) algorithm to simulate the propagation of the ultrasonic waves. The approach is validated here using experiments on AE propagation in thin plate-like structures, where the ultrasound propagates as Lamb waves. Three separate case studies are proposed here. In the first case, an idealised point source is simulated using laser-generated ultrasound, and the optimisation algorithm uses a two-dimensional LISA model. A differential evolution optimisation scheme is used to find the optimal profile of forcing to match the simulation with experiment. In the second case, the two-dimensional LISA approach is used to characterise the forces associated with standard pencil lead breaks. The final study addresses the full three-dimensional wave propagation. Because of the computational expense of the latter calculation, the LISA algorithm is implemented using a CUDA graphics card computer system.