COMPOSITE BLADE CROSS-SECTIONAL DESIGN AND ANALYSIS FOR ADVANCED AIR VEHICLE USING GRAPH NEURAL NETWORK-BASED SURROGATE MODEL

Recently, interest in the AAV (Advanced Air Vehicle) has increased, and research on it has been actively conducted. AAV, a VTOL (Vertical Take-Off and Landing) type aircraft, consists of a rotor hub, blades, etc. Compared to conventional helicopters, AAV blades are shorter, have a large thickness ratio (t/c), and are composed of various composite materials. The general blade cross-section design and analysis process defines the aerodynamic shape, defines internal structure, and then derives the sectional properties (structural and inertial properties) through analysis based on the finite element method. Modal analysis and aeroelastic analysis are performed using a one-dimensional beam model from the derived cross-sectional properties. Moreover, iterative design and analysis are performed to ensure that the derived results satisfy the design requirements. In this process, many design changes occur, and repetitive analyses may induce high computational costs. Therefore, in this study, we will use the deep learning-based reverse engineering surrogate model to find the internal cross-sectional shape that satisfies the given target sectional properties. Tiltrotor example with all cross-sectional properties provided was selected, and the baseline blade was defined by applying Mach scale blade to have a same blade tip speed for the target blade specifications. The baseline blade was subjected to dynamic characteristic analysis and aeroelastic load analysis to satisfy the initial requirements. In the future, we plan to find a cross-sectional internal structure composed of composite materials that satisfies the cross-sectional properties of the baseline blade using the proposed surrogate model and verify it through computational analysis.