Applying to aerodynamic optimization an enhanced particle swarm optimization algorithm based on parallel exchange

基于并行交换的增强粒子群优化算法在气动优化中的应用

School of Aeronautics, Northwestern Polytechnical University, Xi’an 710072, China

Received: 18 June 2021

Abstract

The particle swarm optimization (PSO) algorithm is easy to implement and can obtain high-quality solutions to optimization problems. It is widely applied to nonlinear and difficult problems such as aerodynamic optimization. However, to solve multi-modal problems, it easily falls into locally optimal values, showing that its robustness is poor. In order to improve the robustness of the PSO algorithm, an enhanced particle swarm optimization algorithm based on parallel exchange (EPSOBPE) is proposed. The algorithm enhances the optimization capability and its robustness through the parallel evolution of the cuckoo search algorithm (CSA), PSO population, hierarchical exchange operation and reinforcement learning strategy. Therefore, the algorithm has both the global search capability of the CSA and the local capability of the PSO algorithm, thus making the EPSOBPE very robust. Functional test results show that the EPSOBPE has stronger robustness and adaptability to different problems than other intelligent optimization algorithms. Moreover, the EPSOBPE is applied to the aerodynamic optimization design of the RAE2822 airfoil and the M6 wing. Compared with other algorithms, the EPSOBPE is more robust, and its optimization capability is better.

摘要

粒子群优化(PSO)算法易于实现, 对优化问题可以获得质量较高的解, 被广泛应用在如气动优化这种非线性高难度问题中, 但是面对多峰问题容易陷入局部最优, 存在鲁棒性较差的问题, 为了提高PSO的鲁棒性, 提出了基于并行交换的增强粒子群优化算法(EPSOBPE)。该算法通过布谷鸟搜索算法(CSA)和PSO种群并行进化, 分层交换操作和增强学习策略来增强算法寻优能力与鲁棒性。该算法兼具了CSA的全局搜索能力和PSO的局部能力, 使得新算法具有极强的鲁棒性。函数测试表明, 新算法相较于其他智能优化算法有更强的鲁棒性, 对不同问题的适应能力更强。将EPSOBPE算法应用到RAE2822翼型和M6机翼的气动优化设计中, 相较于其他算法可以得到更好的效果, 从而表明新算法有鲁棒性, 同时兼具了更好的寻优能力。