Optimizing the driving trajectories for guided ultrasonic wave excitation using iterative learning control

摘要

In this work, iterative learning control (ILC) is used to generate precise, guided ultrasonic wave pulses for nondestructive testing (NDT) purposes. The premise of ILC is that the performance of a system, which is repetitively executing the same task, can be improved by learning from previous attempts. In order to develop a systematic approach for the ILC design, an H_∞-synthesis method is employed. To this end, an experimental modal analysis is conducted for commonly used guided ultrasound transducers, and a low-order model is fitted to the resulting data near the transducer's peak frequency. Based on this model, appropriate L- and Q-filters are synthesized and the ILC is trained to generate the desired wave packet in simulation. The simulation results indicate convergence of the proposed ILC algorithm to the optimal driving trajectory. By employing this trajectory in laboratory experiments, it is shown that ringing effects, i.e. distortions of the induced wave packets, are significantly reduced. The ILC is also trained in the experiment, and robustness and convergence are discussed. Lastly, to demonstrate the impact of such an improvement to the induced guided ultrasonic waveform, NDT experiments in the form of edge and delamina-tion detection are performed. It is shown that a significant improvement in reliability can be achieved using the proposed approach. For the MSSP Special Issue in Honor of Professor Lothar Gaul.