An Adaptive Sequential Sampling Strategy-Based Multi-Objective Optimization of Aerodynamic Configuration for a Tandem-Wing UAV via a Surrogate Model

Abstract

Multi-objective optimization of aerodynamic configuration for a tandem-wing unmanned aerial vehicle (UAV) via a surrogate model is appropriate in the primary stages of aircraft design. This study presents an adaptive sequential sampling strategy, which takes into account the principle of entropy rank and selection pooling based on a sigmoid function (ESP), in order to save time and construct a surrogate model database with considerable approximation accuracy. The entire procedure of optimization is divided into four parts, involving problem formulation for design variables and objectives, database construction for the surrogate model, multi-objective optimization with the surrogate models, and ESP adaptive sequential sampling to update the database. Firstly, a comparative study of the different surrogate models is carried out to assess their approximation performance. This verifies that the radial basis function (RBF) surrogate model outperforms the other models across the board. Then, we conduct two tests with typical mathematical problems to validate the effectiveness and applicability of the proposed method. We also develop a multiobjective optimization of the aerodynamic configuration for a tandem-wing UAV, aiming to maximize the lifting coefficient at the ascent (C Lascent ) and the lift-drag ratio (K cruise ) during the cruise. In this case, the RBF surrogate model is proven more suitable than the other common methods to replace the real values calculated by the non-planar vortex-lattice method (VLM) during the process of optimization. Furthermore, a comparison with large minimal distance (LMD) sequential sampling and disposable Latin hypercube sampling (LHS) is carried out alongside the optimization. These results show that the approximation precision achieved using ESP strategy is greater, highlighting the superiority of the ESP adaptive sequential sampling strategy in reducing the number of samples and raising the approximation accuracy. Finally, after the refinement of the database, an optimal Pareto front set is obtained to guide the primary design of the aerodynamic configuration for the tandem-wing UAV. Then, it is verified that the selected trade-off optimal design point has a better aerodynamic performance than the initial reference point, improving C Lascent and K cruise by 6.44% and 10.85%, respectively.

AI-Based Multifidelity Surrogate Models to Develop Next Generation Modular UCAVs

“…In addition, there are surrogate models used in transonic and supersonic wing optimization. Several statistical or machine learning-based surrogate model studies can be found in the literature [38][39][40][41][42][43][44]. When artificial intelligence-based surrogate models are compared with classical methods, it has been observed that as the data set grows, classical methods with statistical approaches become unnecessary and there is the order of magnitude differences in favour of AI methods in terms of computation time [45].…”