Microwave Non-destructive Testing Using K-Medoids Clustering Algorithm

Turbine blades’ metal substrates are often coated with thermal barrier coatings made of composites, notably ceramics. Insulation defects, which might lead to a catastrophic turbine failure, must be detected by routine non-destructive testing. The microwave non-destructive testing has limited spatial imaging, complicating the defect evaluation. This research proposes a unique approach for delamination detection based on microwave non-destructive testing and k-medoids clustering. Using a double-ridged waveguide with 101 frequency points between 18 and 40 GHz, a standard ceramic coating sample is scanned. The k-medoids clustering technique reliably detects and sizes ceramic insulation delamination at each evaluated site. This finding demonstrates the k-medoids clustering method’s capability of detecting delamination with 95.3% accuracy.