【摘要】：響應(yīng)面法是解決橋梁抗震可靠度中非線性和復(fù)雜性等問題的有效方法,但其存在代理模型類型和訓(xùn)練樣本的選取問題.鑒于此,提出了基于樣本填充準(zhǔn)則(ISC)最優(yōu)化和Kriging模型的改進(jìn)序貫抽樣響應(yīng)面法.其以Kriging模型作為代理模型,結(jié)合蒙特卡洛抽樣技術(shù),利用Kriging模型優(yōu)秀的預(yù)測(cè)值估計(jì)性能及獨(dú)有的預(yù)測(cè)均方差估計(jì)能力,建立包含未知樣本預(yù)測(cè)值和預(yù)測(cè)均方差的ISC函數(shù),在迭代階段通過求解ISC最優(yōu)化問題,進(jìn)行局部和全局的平衡搜索,逐一序貫產(chǎn)生后續(xù)訓(xùn)練樣本.最后采用所提的基于ISC-Kriging改進(jìn)響應(yīng)面法對(duì)隨機(jī)地震激勵(lì)下某懸索橋的動(dòng)力可靠度問題進(jìn)行了計(jì)算分析.結(jié)果表明,所提方法具有較高的準(zhǔn)確性、高效性.
[Abstract]:Response surface method is an effective method to solve the nonlinear and complex problems in seismic reliability of bridges, but it has the problem of agent model type and training sample selection. In view of this, an improved sequential sampling response surface method based on sample filling criterion (ISC) optimization and Kriging model is proposed. The Kriging model is used as the proxy model, combined with Monte Carlo sampling technique, and the ISC function with unknown sample prediction value and prediction mean square error is established by using the excellent prediction estimation performance and unique prediction mean square error estimation ability of Kriging model. In the iterative stage, the local and global equilibrium search is carried out by solving the ISC optimization problem, and the subsequent training samples are generated sequentially one by one. Finally, the dynamic reliability of a suspension bridge excited by random earthquake is calculated and analyzed by using the improved response surface method based on ISC-Kriging. The results show that the proposed method has high accuracy and efficiency.
【作者單位】： 華南理工大學(xué)土木與交通學(xué)院;
【基金】：國家自然科學(xué)基金資助項(xiàng)目(51208208) 中國博士后科學(xué)基金資助項(xiàng)目(2013M542174) 華南理工大學(xué)中央高校基本科研業(yè)務(wù)費(fèi)專項(xiàng)資金資助項(xiàng)目(2015ZM114)~~
【分類號(hào)】：U441

【相似文獻(xiàn)】

相關(guān)期刊論文 前3條

1 嚴(yán)根華;閻詩武;駱少澤;陳發(fā)展;;高水頭船閘閥門振動(dòng)動(dòng)力可靠度研究[J];振動(dòng)、測(cè)試與診斷;1996年02期

2 賈布裕;余曉琳;顏全勝;陳舟;;基于Kriging改進(jìn)響應(yīng)面法的橋梁地震動(dòng)力可靠度研究[J];振動(dòng)與沖擊;2013年16期

3 ;[J];;年期

，

本文編號(hào)：2475946