A Dynamic Evolution Method for Digital Twins Based on RDD-RNN

Digital twin technology, as a forefront of research, aims to seamlessly merge physical objects with virtual models, providing new perspectives and solutions to tackle interdisciplinary challenges. Presently, research on algorithms for digital twin models has achieved the accurate prediction of system behavior and performance. Nonetheless, digital twin technology still confronts challenges, such as intricate system modeling and real-time demands. In response, this paper proposes a Real-time Data-Driven Recurrent Neural Network (RDD-RNN) model dynamic evolution approach. By employing data-driven approaches to comprehend the inherent relationships of mechanistic models, the reliance of the twin model on these mechanisms is lessened. Additionally, real-time data is assimilated into the model network in real-time, achieving the dynamic evolution of the twin model. Finally, the effectiveness of the RDD-RNN method was verified through the establishment of a digital twin model for unmanned aerial vehicles and the validation of data related to unmanned equipment flight attitudes control.