Gaussian cross-entropy and organizing intelligence for design optimization of the outrigger system with inclined belt truss in real-size tall buildings实际高层建筑斜向桁架外伸系统设计优化的 Gauss 交叉熵与组织智能
![]()
Farahmand-Tabar S, Ashtari P, Babaei M, 2024. Gaussian cross-entropy and organizing intelligence for design optimization of the outrigger system with inclined belt truss in real-size tall buildings. Probabilistic Engineering Mechanics, 76: 103616.DOI: 10.1016/j.probengmech.2024.103616
摘要 | Abstract
本研究探讨了带有外伸臂和斜向桁架系统的高层建筑最优结构设计。研究采用了 Gauss 交叉熵组织智能 (Gaussian cross-entropy with organizing intelligence, GCE-OI)，一种新的优化方法，利用自组织映射作为机器学习算法，并在交叉熵优化中引入 Gauss 概率分布。这种方法可用于求解方案的前瞻预测，并引导搜索过程快速收敛。优化内容包括构件尺寸 (重量) 和外伸臂的布置 (拓扑)，同时引入斜向桁架，配合传统水平向桁架以提升性能。研究过程给出了风荷载作用下 25 层实际模型的优化，并将结果与多中算法对比。结果表明，所提优化方法在机器学习支持下优于其它算法，提供了更优的求解方案并增强了收敛性。鉴于斜向桁架和所提鲁棒优化方法的有效性，优化布置的外伸系统能够最小化建筑成本并通过限制结构响应来增强结构稳定性。关键词: 外伸系统, 尺寸和拓扑优化, 斜向桁架, Gauss 交叉熵, 机器学习This research explores the optimal structural design for tall buildings with an outrigger and belt truss system. The study employs Gaussian Cross-Entropy with Organizing Intelligence (GCE-OI), a novel optimization approach that utilizes a self-organizing map as a machine learning algorithm, and Gaussian probability distribution in Cross-Entropy optimization. This approach is used to predict promising solutions and to guide the search process for swift convergence. The optimization encompasses member sizing (weight) and outrigger placement (topology) while introducing inclined belt trusses alongside traditional horizontal trusses for enhanced performance. The process involves optimizing a 25-story real-size model subjected to wind load, and the results are compared against multiple well-known algorithms. The results show that the proposed optimizer, supported by machine learning, outperforms alternative algorithms, offering superior solutions with enhanced convergence. Considering the efficiency of the inclined belt trusses and the proposed robust optimization method (GCE-OI), the optimally-placed outrigger system minimizes the constructional cost and enhances structural stability by limiting the responses.Keywords: Outrigger system; Size and topology optimization; Inclined belt truss; Gaussian cross-entropy; Machine learning创新点 | Highlights
- Innovative cross-entropy optimization with Gaussian probability distribution
- Organizing intelligence for data clustering and prediction of promising solutions
- Improving the performance of the outrigger system with inclined belt truss
- Simultaneous size and topology optimization for structural efficiency
- Better efficiency of ML-driven Gaussian cross-entropy optimization among competitors
![]()
Fig. 1. Interaction of the outrigger and the core system
图 2: 外伸系统中所提与传统支撑桁架示意图: (a) 水平向; (b) 斜向Fig. 2. Schematic view of proposed and conventional belt truss in outrigger system: (a) Horizontal; (b) Inclined
图 3: 外伸系统的简单斜向桁架: (a) 水平向; (b) 斜向Fig. 3. Simple and inclined belt truss in outrigger system: (a) Horizontal; (b) Inclined
图 4: 典型模型的设计分组: (a) 立面图; (b) 平面图Fig. 4. Design groupings in a typical model: (a) Elevation; (b) Plan
Fig. 5. Self-organizing map
图 6: 自组织映射中聚类与数据减缩的重要性: (a) 训练函数; (b) 云图; (c) 聚类与数据减缩Fig. 6. Significance of clustering and data reduction in SOMs: (a) Training function; (b) Contour illustration; (c) Clustering and data-reduction (lattice of 10 × 10 neural units)
Fig. 7. Competitive layer in 10x10 network of neural units
图 8: 具有组织智能的 Gauss 交叉熵优化流程图Fig. 8. Flowchart of Gaussian cross-entropy optimization with organizing intelligence
图 9: 采用水平向桁架系统下建筑的收敛时程与层间位移Fig. 9. Convergence history and inter-story drift of the building with the horizontal truss belt system
图 10: 采用斜向桁架系统的建筑收敛时程与层间位移Fig. 10. Convergence history and inter-story drift of the building with the inclined truss belt system
作者信息 | Authors
Salar Farahmand-Tabar, 共同通讯作者 (Corresp.) 伊朗赞詹大学 (University of Zanjan) 工学院Email: farahmandsalar@znu.ac.ir
Payam Ashtari, 共同通讯作者 (Corresp.) 伊朗赞詹大学 (University of Zanjan) 工学院Email: ashtari@znu.ac.ir
伊朗赞詹大学 (University of Zanjan) 工学院
律梦泽 M.Z. Lyu | 编辑 (Ed)
P.D. Spanos | 审校 (Rev)
陈建兵 J.B. Chen | 审校 (Rev)
彭勇波 Y.B. Peng | 审校 (Rev)
