A framework for the performance evaluation of reconfigurable ultrasonic sparse array networks

Autonomous monitoring strategies are becoming an increasingly popular application modality due to their adaptable deployment and reconfigurability. Motivated by low-cost robotic swarms conveying ultrasonic transducers that are maneuverable into arrays for pipe inspections, this paper aims to develop an initial framework for predicting such an array’s performance and understanding the effects of target defect scattering properties and array topology parameters on that performance. Assuming uniformly omnidirectional transducers, we first develop a predictive spatial probability of detection (POD) model that depends on target defect scattering and distance and verify it against experiment. We then synthesize these single-transducer models into arrays and derive global POD performance metrics that are parameterized by defect scattering properties (via the single transducer model) and array topology (geometric arrangement and transducer pitch). We perform a performance evaluation in the case of a highly directional scatterer (e.g., a crack) over the global parameter space to make suggestions about array design.