A team of researchers in the Army’s research lab has developed and tested networked acoustic sensors that detect airframe damage on composite UH-60 Black Hawk helicopters.
The Army has been studying several possible ways to monitor the integrity of helicopter airframes for about two years as it looks for a way to reliably detect and locate in real time the initiation and spread of damage inflicted during operations. The team had considered ultrasonics and radiography for this purpose, but rejected them: each requires an external energy source to create a directed wave. The external energy would also interfere with other aircraft subsystems.
Acoustic sensing, however, is a passive and non-destructive technique. It can detect damage in the early stages, long before there is any catastrophic damage or failure. And unlike other methods, acoustics sensing detects damage in real time (or at the instant the damage is being done).
The ARL sensing network is composed of several lightweight transducers encapsulated in 3D-printed sensor mounts. The sensors are distributed in several damage-prone areas zones to widen coverage and improve probability of damage detection. Data acquisition uses a controllable timing parameter to reject reflections of direct waves and waves coming from non-damage related events.
The signal processing algorithm includes a layer of adaptive filters to reduce the effects of signal distortion during location analysis. In a test, the sensor network consistently identified and located the initiation and progress of damage during a fatigue test that lasted over 200,000 cycles.
Currently, the Army keeps its aircraft flying by using phase maintenance, which is periodic calendar-based inspections and component replacement. The process is inefficient, costly, and entails extended downtime, but it does keep aircrews safe. The newly developed sensor network will let the Army conduct condition-based maintenance and drastically cut the life cycle cost of Army vehicles. The work also supports the Army’s long-term vision of maintenance-free aircrafts.
Large-scale acoustic monitoring is data-intensive, with several million hits being received by each transducer per flight. This puts a high load on the internal bus and circuitry of data acquisition hardware. In general, most hits are not related to damage. Rather, they are noises from moving parts such as the clicking or rubbing noises of fasteners, panel connections, and vibrations from other non-damage related sources.
Unwanted acoustic hits also arise from reflections of an already received and processed sound. The challenge is to develop a network sensitive only to damage-related hits and insensitive to all others.