Data fusion for automated non-destructive inspection.

Area of Science:

• Industrial Engineering
• Materials Science
• Computer Science

Background:

• Automated non-destructive evaluation (NDE) generates large datasets requiring manual analysis.
• Multiple data acquisitions in NDE offer potential for improved inspection reliability.
• Current manual analysis methods cannot systematically combine diverse NDE readings.

Purpose of the Study:

• To develop a data-fusion software framework for partial automation of NDE data analysis.
• To optimize operator time by reliably identifying defect-free regions and defect indications.
• To enhance the reliability of automated NDE through systematic data combination.

Main Methods:

• Development of a data-fusion based software framework.
• Application of the framework to industrial ultrasonic immersion inspection of aerospace turbine discs.
• Processing of industrial datasets to evaluate system performance.

Main Results:

• Demonstrated an orders-of-magnitude reduction in false-call rates for a given probability of detection.
• Achieved high probability of declaring component regions defect-free.
• Successfully identified defect indications, optimizing operator focus.

Conclusions:

• The data-fusion framework offers a significant improvement in NDE inspection reliability and efficiency.
• The system is applicable to various automated NDE scenarios, exemplified by aerospace applications.
• Partial automation through data fusion effectively addresses the challenges of analyzing large NDE datasets.