Ultrasonic NDT optimization using Randomized Adaptive Differential Evolution
Abstract
Ultrasonic non-destructive testing (UT) enables to control the integrity of critical parts. The design of transducers and the optimization of inspection procedures might be involved tasks due to the possibly large number of parameters to consider. For complex configurations, experts generally have to perform a parametric study using simulation tools, which can be time-consuming and fastidious. Here, an automatic method for the optimization of inspection parameters (positioning, setting and/or designing the transducer) is proposed. It uses an evolutionary algorithm linked with forward modelling algorithms that are already implemented in the ultrasonic module of CIVA. The method is then successfully tested on a realistic UT application of nozzle inspection, and further extended to handle constraints and achieve specific accuracy in order to optimize the design of a transducer.