岩质岸坡岩样力学参数取值方法比较

针对安川公路典型岸坡岩石试样, 分别在天然状态和饱水状态下对其进行轴压试验, 采用随机模糊法、最小二乘法和蒙特卡罗法对岩石变形参数和抗剪强度参数进行抽样取值, 并计算了相应的边坡稳定性系数。计算结果表明: 采用蒙特卡罗法抽样得到的变形参数均值、均方差与变异系数与最小二乘法基本接近, 其中均值相差仅为0.77%;采用随机模糊法抽样得到的变形参数均值、均方差与变异系数最小, 其均值较最小二乘法降低了8.60%;在抗剪强度参数抽样取值时, 利用随机模糊法得到的抗剪强度参数均值、变异系数与对应的边坡安全系数最小, 蒙特卡罗法次之, 最小二乘法最大, 其中采用随机模糊法抽样取值时边坡安全系数均值为1.243, 蒙特卡罗法取值时为1.521。相比于蒙特卡罗法, 采用随机模糊法确定的隶属度函数考虑了样本参数权重因素的影响, 反映了岸坡岩石抗剪强度参数的不确定性, 计算的岩质边坡安全系数更具有工程指导意义。

Comparison of selecting methods on rock sample mechanical parameters of rock bank slope

The axial compression tests of rock samples of typical bank slope in Anchuan Highway were carried out under natural state and water-saturated state, their deformation parameter and shear strength parameters were sampled and selected by the random-fuzzy method, the least square method and the Monte Carlo method, and the stability coefficient of typical bank slope was computed.Calculation result shows that the mean, standard deviation and variation coefficient of deformation parameter obtained by the Monte Carlo method are basically similar to the values obtained by the least square method, and the mean difference is only 0.77%.Relatively, the mean, standard deviation and variation coefficient of deformation parameter obtained by the random-fuzzy method are minimal, and the mean value is 8.60%lower than the value obtained by the least square method.In sampling and valuing of shear strength parameters, the means and variation coefficients of shear strength parameters and the corresponding safety factors of slope from small to large are obtained in turn by the random-fuzzy method, the Monte Carlo method and the least square method, and slope safety factors are 1.243 and 1.521 when the parametersare computed by the random-fuzzy method and the Monte Carlo method respectively.Compared with the Monte Carlo method, the membership function confirmed by the random-fuzzy method considers the weight effect of sample parameter, which can reflect the uncertainty of rock shear strength parameter.The safety factors of rock slope calculated by the random-fuzzy method have more engineering guiding significance.