摘要
多学科优化设计(MDO)是当前复杂系统工程设计中研究最活跃的领域。分析了标准多学科协同优化算法解决实际复杂MDO问题计算困难的原因,提出了基于试验设计的近似模型和智能优化的协同优化算法(NCO)。NCO算法继承了标准协同优化分布并行的思想,采用现代智能算法优化系统级减小优化陷入局部解的可能性,以试验设计为基础的高精度近似模型代替学科真实模型降低计算成本,平滑数值噪声。通过经典MDO测试算例与Alexandrov提出的改进松弛协同优化比较,优化结果表明,NCO能有效提高收敛速率,保证收敛结果的稳定性和可靠性,能更好地满足复杂系统工程优化需要。
MDO（multidisciplinary design optimization） is the most active research area in current complex system engineering.Reasons of the defects of traditional collaborative optimization to solve the complex multidisciplinary are analyzed and a new collaborative optimization based on approximation models and intelligent optimization methods is proposed.Intelligent optimization methods help to reduce the possibility of falling into a local solution.Meanwhile,high precision approximation models relied on design of experiments also improve convergence rate and smooth the numerical noise.Classic example is adopted to test the new collaborative optimization.Results shows that the new collaborative optimization can effectively improve the rate of convergence,the stability and reliability of the optimization results.Meanwhile,the presented method is better to meet the needs of the complex system engineering.
出处
《计算机与数字工程》
2014年第1期65-68,116,共5页
Computer & Digital Engineering
关键词
协同优化
智能算法
试验设计
近似模型
collaborate optimization
intelligent optimization
design of experiments
approximation models
