| Popis: |
Sensor data validation is a key module in any fault diagnosis system. A pattern recognition based algorithm is therefore proposed to validate the reliability of an arbitrarily distributed sensor network and its data. Single-frequency Lamb wave experiments are conducted on aircraft qualified aluminium plates, which have perfectly bonded and debonded Lead Zirconate Titanate patches along with structural damage in the form of holes. Sensor debondings are examined using sensor network data. Simple features like standard deviation, autocorrelation are extracted from Lamb wave signals and employed in the pattern recognition system that distinctly identifies the sensor debonding albeit the structural damage is present. |
