Today’s military and commercial aircraft consist of millions of structural, hydraulic, and computer components housed in a strong metallic and composite body. To inspect and repair all these components, planes and rotorcraft must be opened or disassembled to access every piece. Because these machines require precise manufacturing in order to fly, small filaments, wiring, and even dust can affect aircraft componentry. Leveraging advanced technology can enable non-destructive evaluation (NDE) so that vehicles do not have to be opened, cut, or disassembled to complete a full inspection.
To maintain its air domain advantage, the United States Air Force requested a handheld NDE tools to detect and identify small foreign object debris (FOD) that were invisible to the naked eye. This kind of debris appears most often during the manufacturing process, when small filaments of metal and carbon fiber can spread across and inside key components. They needed an advanced sensor to detect these filaments and ensure that they are cleaned off the aircraft’s parts.
To meet this need, Intellisense devised a compact, handheld scanner that combined optical and terahertz (THz) inspection cameras with anomaly-detection software to automatically find these small filaments. This innovative integration enabled inspectors to review the full width of a thin mat manufacturing line (which is generally wider than 54 in.) through during real-time workflow (which is generally at a speed of 13 ft per minute or higher). It can automatically detect and record defects as well as nonvisual FOD to improve materials manufacturing and quality assurance processes.
The development process began with the integration of radio frequency scattering simulation and algorithms for THz imagery processing and FOD detection. Through testing, Intellisense demonstrated the feasibility of the THz scanning approach by detecting conductive FOD within prepreg mats (i.e., a reinforcing fabric that has been pre-impregnated with a resin system). The THz scattering simulation was able to detect FOD with a radius of 100 micrometers, which are practically invisible to the naked eye. It also achieved a scan width of 54 in. and speed exceeding the Air Force’s 13-feet-per-minute requirement. The system offers both visual and nonvisual coverage, as well as a high sensitivity and full detection automation to handle different mat materials and manufacturing environments.
Detecting FOD within micrometers of resolution required complex deep-learning algorithms to be integrated into a small, handheld sensor. Although high-resolution visibility cameras were more amenable to this compact packaging, the team at Intellisense found the deep-learning algorithms were more beneficial because it could be improved with training from datasets. This software could also generate a visual or audio cue to the operator for immediate human intervention when any FOD is detected.
Thanks to the integration of optical and THz sensors with deep-learning software, inspectors and crewmates can evaluate the interiors of sophisticated equipment without having to make any openings, cuts, or breakages. The small form factor also enables manufacturers to use or install the device in a variety of settings and circumstances. Ultimately, the flexibility of this non-destructive testing and evaluation tool ensures that Air Force planes are free of defects and can fly safely without long and laborious inspection processes. These kinds of innovations ultimately save the Air Force time and millions in taxpayer money.