人民黄河
人民黃河
인민황하
Yellow River
2014年
5期
116-119
,共4页
初始地应力%回归分析%人工神经网络%锦屏水电站
初始地應力%迴歸分析%人工神經網絡%錦屏水電站
초시지응력%회귀분석%인공신경망락%금병수전참
initial geostress%regression analysis%artificial neural network%Jinping Hydropower Station
结合锦屏二级水电站复杂的地质条件,运用有限元法和BP人工神经网络对研究区的初始地应力场进行了反演分析。结果表明:主应力的变化规律为从上到下逐渐增大,主应力的分布随埋深的增加而增大,主应力等值线在浅层区域受地形起伏影响很大;在靠近地表及河谷底部,等值线分布较密集,应力变化梯度较大,出现应力集中现象;地下厂房区域表层及浅层主应力倾角与山体坡角基本相同,在深部区域,最大主应力倾角随深度的增加逐渐增大;数值模拟反演计算得到的初始应力场分布与研究区的地形地貌关系密切,而岩性对地应力场的分布影响较小;BP人工神经网络反演值与实测值较接近,反演精度可以满足工程实际需要。
結閤錦屏二級水電站複雜的地質條件,運用有限元法和BP人工神經網絡對研究區的初始地應力場進行瞭反縯分析。結果錶明:主應力的變化規律為從上到下逐漸增大,主應力的分佈隨埋深的增加而增大,主應力等值線在淺層區域受地形起伏影響很大;在靠近地錶及河穀底部,等值線分佈較密集,應力變化梯度較大,齣現應力集中現象;地下廠房區域錶層及淺層主應力傾角與山體坡角基本相同,在深部區域,最大主應力傾角隨深度的增加逐漸增大;數值模擬反縯計算得到的初始應力場分佈與研究區的地形地貌關繫密切,而巖性對地應力場的分佈影響較小;BP人工神經網絡反縯值與實測值較接近,反縯精度可以滿足工程實際需要。
결합금병이급수전참복잡적지질조건,운용유한원법화BP인공신경망락대연구구적초시지응력장진행료반연분석。결과표명:주응력적변화규률위종상도하축점증대,주응력적분포수매심적증가이증대,주응력등치선재천층구역수지형기복영향흔대;재고근지표급하곡저부,등치선분포교밀집,응력변화제도교대,출현응력집중현상;지하엄방구역표층급천층주응력경각여산체파각기본상동,재심부구역,최대주응력경각수심도적증가축점증대;수치모의반연계산득도적초시응력장분포여연구구적지형지모관계밀절,이암성대지응력장적분포영향교소;BP인공신경망락반연치여실측치교접근,반연정도가이만족공정실제수요。
Combined with complex geological conditions of Jinping Hydropower Station,the finite element method and BP artificial neural network had been used for regression analysis of initial geostress field in research area. The results show that the principal stress increases gradually from top to bottom,the distribution of principal stress increases with the increasing buried depth,the main stress contour is greatly affected by topography in the shallow area;close to the surface and the bottom of the valley,the distribution of stress contour is more intensive,the change gradient of stress is greater and stress concentration phenomenon appeared;the surface and shallow principal stress dip angle is basically the same as the mountain slope angle in the underground workshop area,in the deep region,the maximum principal stress inclination increased with the increasing of depth;the initial stress field distribution calculated by numerical simulation is closely related to the topography in the study area,while the lithology has smaller influence on the distribution of stress field;stress measured value is closer to the regression value of BP artificial neural network and regres-sion accuracy can meet the actual needs of the project.