Model Updating Method for Digital Twin of Unmanned Aerial Vehicle Based on Bayesian Inference and Improved Pigeon-Inspired Algorithm

In this paper, a method for updating the digital twin models of Unmanned Aerial Vehicles (UAVs) is proposed. The accuracy and utility of digital twins models depend on their ability to reflect the current state and behavior of the physical UAVs with high fidelity. The method proposed aims to update the model parameters through Bayesian Inference. The maximum posterior estimation is solved by an improved PIO that incorporates predator behavior, chaotic mapping, and a Trap-Avoidance Operator (TAO) to improve the efficiency of the search process for the maximum posterior estimation. In addition, an experimental validation through online sampling and parameter updating during UAV flight has demonstrated that the proposed method can update UAV model parameters while maintaining high computational efficiency and accuracy. It has significance for the real-time synchronization and optimization of UAV digital twin models. Furthermore, a comparison between the improved PIO and other optimization algorithms shows that the improved PIO algorithm has the advantage in initial adaptability and convergence speed, illustrating its strength in updating digital twin models of UAVs.