施工技术
施工技術
시공기술
CONSTRUCTION TECHNOLOGY
2014年
11期
112-114
,共3页
地基%堆载预压%沉降%软土%预测%数值模拟
地基%堆載預壓%沉降%軟土%預測%數值模擬
지기%퇴재예압%침강%연토%예측%수치모의
foundations%surcharge preloading%settlement%soft soil%prediction%simulation
针对软土地基含水量高、压缩性高及抗剪强度低的特点,采用理论分析与数值模拟相结合的方法,从力学机理探讨了堆载效应与软土地基沉降的关系。软件FLAC3D数值模拟表明,随着地基深度的增加,地基的竖向沉降不断减少,并且其递减幅度随深度的增加而增大,最大水平向位移发生在堆载物边缘以下的深部地基土中;随着水平距离的增加,地基竖向沉降也不断减少,并且其减少的幅度较均匀,当水平距离增加到一定值,其沉降值维持在某一个恒定值。以现场监测资料为基础构建动态模糊神经网络,获得了准确度和可靠度较高的预测结果,表明动态模糊神经网络具有较强的网络稳定性和泛化能力。
針對軟土地基含水量高、壓縮性高及抗剪彊度低的特點,採用理論分析與數值模擬相結閤的方法,從力學機理探討瞭堆載效應與軟土地基沉降的關繫。軟件FLAC3D數值模擬錶明,隨著地基深度的增加,地基的豎嚮沉降不斷減少,併且其遞減幅度隨深度的增加而增大,最大水平嚮位移髮生在堆載物邊緣以下的深部地基土中;隨著水平距離的增加,地基豎嚮沉降也不斷減少,併且其減少的幅度較均勻,噹水平距離增加到一定值,其沉降值維持在某一箇恆定值。以現場鑑測資料為基礎構建動態模糊神經網絡,穫得瞭準確度和可靠度較高的預測結果,錶明動態模糊神經網絡具有較彊的網絡穩定性和汎化能力。
침대연토지기함수량고、압축성고급항전강도저적특점,채용이론분석여수치모의상결합적방법,종역학궤리탐토료퇴재효응여연토지기침강적관계。연건FLAC3D수치모의표명,수착지기심도적증가,지기적수향침강불단감소,병차기체감폭도수심도적증가이증대,최대수평향위이발생재퇴재물변연이하적심부지기토중;수착수평거리적증가,지기수향침강야불단감소,병차기감소적폭도교균균,당수평거리증가도일정치,기침강치유지재모일개항정치。이현장감측자료위기출구건동태모호신경망락,획득료준학도화가고도교고적예측결과,표명동태모호신경망락구유교강적망락은정성화범화능력。
Based on high water content, high compressibility and low shear strength characteristics in soft soil foundation, combined with theoretical analysis and numerical simulation, this paper introduces the relationship between surcharge effect and settlement in mechanical mechanism respect in soft soil foundation. Software FLAC3D numerical simulation shows that the vertical foundation settlement decreases gradually with the foundation depth increased, and the decreasing rate increases with depth increased. The maximum horizontal displacement occurs in deep foundation soil under the edge of the surcharge object. The vertical settlement also decreases with the horizontal distance increased, and the decreasing rate is relatively uniform, a constant settlement is achieved at certain horizontal distance. Based on field monitoring data,the dynamic fuzzy neural network ( DFNN) is constructed,high accuracy and reliable prediction results are achieved. The prediction results show that the dynamic fuzzy neural network is a stable network with high prediction ability.