ECCOMAS 2024

Optimization Design of Aircraft Structure Based on Strain Neighborhood Genetic Algorithm

  • ZHANG, Rui (CAE)
  • Cui, Degang (S&T Committee of AVIC)
  • Wang, Kaijian (CAE)

Please login to view abstract download link

When the stiffness of composite materials varies in space, the tailoring and optimization of structure performance can be achieved. For high aspect ratio wing, the aerodynamic load of the wings is gradually transmitted from the tip through the skin,rib,spar to the root. Due to the design constraints of wing loading and leading and trailing edge control surfaces, the layout parameters of the wing beam vary less. Based on the weight proportion of the wing spar, rib and skin, the skin area accounts for a larger proportion. Therefore, optimization of the skin stiffness can achieve significant improvement space and wegiht reduction.Large scale design variable optimization expands the design space and can achieve more refined optimization design. However, at the same time, the design scheme has poor processability due to the lack of constraints on the process during design. In particular, large-scale design variables often use the global optimization algorithm, and the optimization results are decentralized, unable to meet the requirements of manufacture, and can not achieve the continuity of force transmission. This article proposes a genetic algorithm based on strain neighborhood and applies it to the structural optimization of composite material wings, achieving a reduction in structural weight while reducing sudden changes in skin ply thickness. The strain neighborhood based method proposed in this article considers the strain of the upper and lower wing finite element elements as a dataset. In these datasets, several higher strains will appear, making these elements cluster centers. Based on the position of the element where these higher strains are located as the center, according to the principle of four connections, the element with monotonic changes in strain values and common edges will be divided into a "neighborhood cluster", multiple neighborhoods can be established based on the strain range. The genetic algorithm optimization is based on the "neighborhood cluster" as the object. The units within the cluster are optimized as a whole, reducing the number of design variables and thickness variations between adjacent units, and also reflecting the continuity of force transmission, improving the continuity and smoothness of the skin laminate.