昆明理工大学学报（自然科学版）
昆明理工大學學報（自然科學版）
곤명리공대학학보（자연과학판）
JOURNAL OF KUNMING UNIVERSITY OF SCIENCE AND TECHNOLOGY（SCIENCE AND TECHNOLOGY）
2014年
4期
43-47
,共5页
曹净%丁文云%赵党书%宋志刚%桂跃
曹淨%丁文雲%趙黨書%宋誌剛%桂躍
조정%정문운%조당서%송지강%계약
基坑支护结构%响应面法%LS-SVM%PSO%均匀试验%优化设计
基坑支護結構%響應麵法%LS-SVM%PSO%均勻試驗%優化設計
기갱지호결구%향응면법%LS-SVM%PSO%균균시험%우화설계
foundation pit supporting structure%response surface method%LS-SVM%PSO%uniform design%optimization design
由于受安全性和经济性的双重指标要求，基坑支护结构优化设计一直是基坑工程中的研究重点之一。本文结合均匀试验设计（UD）和粒子群算法优化的最小二乘支持向量机（PSO-LSSVM），提出了一种基于LS-SVM响应面法的基坑支护结构优化设计方法。其中，以工程量为优化目标，以安全系数作为约束条件，首先通过UD和有限元计算高效构造学习样本，然后利用PSO-LSSVM模型的高度非线性映射能力建立学习样本响应面，并预测由Monte-Carlo 模拟生成的多组随机样本结果，最后从中筛选满足全部约束条件的且目标函数值最小的样本作为优化设计最优解。通过算例分析表明，该方法优化效果明显，且具有较好的效率和精度。
由于受安全性和經濟性的雙重指標要求，基坑支護結構優化設計一直是基坑工程中的研究重點之一。本文結閤均勻試驗設計（UD）和粒子群算法優化的最小二乘支持嚮量機（PSO-LSSVM），提齣瞭一種基于LS-SVM響應麵法的基坑支護結構優化設計方法。其中，以工程量為優化目標，以安全繫數作為約束條件，首先通過UD和有限元計算高效構造學習樣本，然後利用PSO-LSSVM模型的高度非線性映射能力建立學習樣本響應麵，併預測由Monte-Carlo 模擬生成的多組隨機樣本結果，最後從中篩選滿足全部約束條件的且目標函數值最小的樣本作為優化設計最優解。通過算例分析錶明，該方法優化效果明顯，且具有較好的效率和精度。
유우수안전성화경제성적쌍중지표요구，기갱지호결구우화설계일직시기갱공정중적연구중점지일。본문결합균균시험설계（UD）화입자군산법우화적최소이승지지향량궤（PSO-LSSVM），제출료일충기우LS-SVM향응면법적기갱지호결구우화설계방법。기중，이공정량위우화목표，이안전계수작위약속조건，수선통과UD화유한원계산고효구조학습양본，연후이용PSO-LSSVM모형적고도비선성영사능력건립학습양본향응면，병예측유Monte-Carlo 모의생성적다조수궤양본결과，최후종중사선만족전부약속조건적차목표함수치최소적양본작위우화설계최우해。통과산례분석표명，해방법우화효과명현，차구유교호적효솔화정도。
Due to the requirement of safety and economy,the optimization design of foundation pit supporting structure has been one of the research emphasis in pit engineering.The uniform design (UD)and least square support vector machine optimized by particle swarm algorithm (PSO-LSSVM) are combined to constitute a response surface method.The method is used to optimize the supporting structure of foundation pit.In this paper,the engineering quantity is regarded as optimization objective and safety factors are regarded as restrictive conditions.The UD and finite element calculation are used to construct study samples efficiently.The capability of nonlinear mapping of PSO-LSSVM is employed to build the response surface model.On this basis,multiple sets of random samples generated by Monte-Carlo simulation are predicted by the model.Finally,a sample with the minimum value of objective function with its output indicators that satisfy all constraints is screened out as the optimal solution.An engineering example is studied and the result shows that this method can obtain obvious optimization effect with better efficiency and accuracy.
