The use of unmanned aerial vehicles (UAVs) for structural health inspection has become a promising technique to perform labor-intensive, accessibility-challenged, and sometimes dangerous inspection tasks. This paper presents a novel physics-informed UAV inspection planning framework for infrastructure structural health assessment based on model-based diagnostics and prognostics enabled by physics-based probabilistic analysis. It bridges the gap between UAV mission planning and inspection with model-based probabilistic analysis, by allowing bidirectional information exchange, namely (1) structural damage state diagnostics using UAV inspection data and (2) UAV inspection optimization through model-based failure prognostics. Based on the bidirectional communication, the impacts of the three key UAV inspection parameters (i.e., inspection distance, inspection interval, and critical maintenance threshold) on structural life-cycle cost are analyzed. The optimization of key UAV inspection parameters is achieved by minimizing the cost per unit time (CPUT) through model-based pre-posterior analysis. In this analysis, synthetic observations are generated using predictive models according to the prior distributions of various uncertainty sources (e.g., detection rate, damage state evolution, etc.). The generated synthetic observations are then used to obtain the posterior distributions of uncertain parameters, enabling the integration of prior information and Bayesian model updating into inspection optimization through a cost function. The proposed model offers a robust method to accommodate the inherent uncertainties in failure prognostics, leading to a more effective optimization of the UAV inspection parameters. The practical application of the framework is demonstrated through a miter gate example. The results show that the proposed method is able to efficiently determine the optimal UAV inspection parameters and continuously update the information model.
Read full abstract