Performance Modeling and Trajectory Optimization of eVTOL Using RBF Neural Network

Abstract Electric Vertical Take-Off and Landing (eVTOL) is a key development direction for the future of electric urban air mobility (UAM),which are characterized by multiple lift surfaces, control surfaces,and operational modes.eVTOL offers superior flight capabilities, enabling high-speed travel while maintaining vertical takeoff and landing capabilities. However,the inherent complexity of the aircraft’s dynamic model complicates its solution, thereby limiting its applicability in trajectory optimization. This paper presents a dynamic model of the aircraft for analyzing its flight performance. A Radial Basis Function (RBF) neural network surrogate model is subsequently employed to construct the aircraft performance model. The effectiveness of the model is validated through RBF neural network-based fitting analysis.A case study on urban low-altitude flight trajectory optimization further demonstrates the advantages of the flight performance surrogate model:it effectively captures the capabilities of eVTOL and maximizes the utilization of the aircraft’s performance limits.