Unmanned Aerial Vehicles (UAVs) are receiving increasing attention for use in Non-Destructive Testing / Evaluation (NDT / NDE), due to affordability, safety, and the ability to access areas where manned inspection is not practical. The mobility and size of UAVs offer the flexibility to quickly deploy remote inspections of large-scale assets and targets with complex geometry, such as wind turbine blades. The platform also provides airborne approaches to inspect high-risk sites, such as areas within nuclear facilities, which conventionally require inspectors to work at high altitude or in the presence of safety risks. Although ground crawler vehicles have been seen in such applications, UAVs offers true three-dimensional (3D) inspections and solve many challenging access problems. Commercial UAV-deployed NDT inspections typically rely on a high-resolution camera that is manually piloted with a relatively large standoff distance, due to collision concerns and aerodynamic challenges. Close-range and contact-based inspections grant more detailed and accurate evaluations, whilst demanding an advanced UAV flight control system. This thesis evaluates two approaches to automated remote NDT, namely 3D photogrammetric and ultrasonic inspection. 3D photogrammetric inspection is a visual NDT method used to quantify surface integrity and detect discontinuities in a structure's coating. Compared with traditional inspections from offline images, a 3D photogrammetric inspection provides results with intrinsic position and location information, which is essential to meaningful surface condition evaluations. Structural health conditions, such as internal support material corrosion and fatigue crack formation beneath an outersurface coating, require contact-based measurement technologies. Contact-based ultrasonic measurements grant the opportunity to remotely monitor the structural health of an industrial asset with enhanced internal integrity information. This thesis has investigated and evaluated photogrammetric and contact-based inspections deployed by an aerial vehicle. The novel contribution of the research is to investigate the connection between the inspection accuracy and parameters in UAV-deployed inspections. Additionally, a novel, airborne, ultrasonic measurement system for the contact inspection of non-magnetic facilities was established. As far as the author aware, this is the first time such an inspection has been implemented and evaluated. The implementation and modification of the autonomous flight controller on an AscTec Firefly UAV for indoor inspections are detailed within. A miniature planar laser scanner with a curve fitting was integrated into the system to map the UAV surroundings, measuring the displacement and alignment error against the inspection target. Brightness conditions, motion blur, and focal blur parameters influenced the accuracy of the photogrammetric inspections and were quantified and discussed. The inspection accuracy was improved from 0.3853 mm to 0.3098mm using more detailed features on the mesh while the photogrammetric inspection was undertaken with a laser-based flight trajectory. The negative influences of these parameters were mitigated with an active flight path and an appropriate experimental setup, improving the reconstruction error by a factor of 13 versus the poorest scenario. An autonomous UAV-deployed ultrasonic measurement system was developed for the inspection of vertical structures. The inspection constraints, such as transducer alignment and UAV positional accuracy, impacted the measurement error during the ultrasonic inspections. The influences of these constraints were evaluated and analysed. An additional contribution was presented on the improved inspection accuracy while applying a coded excitation to the ultrasonic measurement system to increase Signal-to-Noise Ratio. The maximum error was reduced from 0.83 mmto 0.66 mm in scenarios where the transducer was sub-optimally aligned.
|Date of Award||31 May 2019|
- University Of Strathclyde
|Sponsors||University of Strathclyde|
|Supervisor||Gordon Dobie (Supervisor) & Anthony Gachagan (Supervisor)|