基于高斯过程的萤火虫算法及其在板料成形优化设计中的应用

发布时间：2018-01-20 16:37

本文关键词： 高斯过程近似模型萤火虫算法板料成形　出处：《锻压技术》2015年12期 　论文类型：期刊论文

【摘要】：在板料成形优化设计等领域中,许多复杂工程优化问题往往表现出高维以及多重非线性耦合性等特征,从而导致此类问题优化收敛效率大幅下降。本文提出了一种基于高斯过程近似模型的高维全局优化方法,此方法最大优势在于:能够建立基于高斯过程近似模型的搜索机制,利用该机制并结合EI(Expected Improvement)准则,对萤火虫算法生成的搜索训练样本进行筛选,从而自动产生新的样本点。在此模式下进行优化迭代能够确保算法的搜索快速集中在全局最优点的较小区域内,进而大幅提升优化效率的同时确保收敛的稳健性。复杂的高维非线性测试函数表明,算法具有处理复杂高维工程问题的能力。此外,算法成功应用于NUMISHEET2013 BENCHMARK2中盒形件的板料成形中,对基于时变的压边力进行了优化设计。同主流的EGO(Efficient Global Optimization)相比,其精度和效率都得到了明显的提升。
[Abstract]:In the field of sheet metal forming optimization design, many complex engineering optimization problems often show the characteristics of high dimension and multiple nonlinear coupling. As a result, the convergence efficiency of this kind of optimization problem is greatly reduced. In this paper, a high-dimensional global optimization method based on Gao Si process approximation model is proposed. The biggest advantage of this method lies in its ability to establish a search mechanism based on Gao Si's process approximation model, which can be used in combination with the EI(Expected implementation criterion. The search training samples generated by the firefly algorithm are filtered to automatically generate new sample points. In this mode, the optimization iteration can ensure that the search of the algorithm is quickly concentrated in the smaller region of the global best. Moreover, the complex high-dimensional nonlinear test function shows that the algorithm has the ability to deal with complex high-dimensional engineering problems. The algorithm is successfully applied to sheet metal forming of box parts in NUMISHEET2013 BENCHMARK2. The optimized design of the beam-holding force based on time-varying is carried out, which is compared with that of the mainstream EGO(Efficient Global optimization. Its precision and efficiency are obviously improved.
【作者单位】：湖南大学汽车车身先进设计制造国家重点实验室;
【基金】：国家自然科学基金资助项目(11172097,11302266)
【分类号】：TG386;TP18
【正文快照】： 随着计算机技术和数值计算方法的发展,有限元模拟技术逐渐成为板料成形分析中的重要手段并广泛应用于板料成形优化设计中[1]。与传统的物理实验和试错法相比,有限元仿真模拟技术的应用大幅降低了板料成形优化设计中的时间和成本。然而,在工程实践应用中,对结果的判断和相关参

【参考文献】