黄土广泛分布于我国西北地区，伴随季节性的冻结和融化，黄土的力学特性极易发生改变，甚至引发工程问题，如路基冻胀、融沉、翻浆冒泥等。本文以西北地区黄土为研究对象，采用水泥和聚丙烯纤维进行黄土改性，研究水泥掺量、纤维长度、纤维掺量以及冻融循环次数对改性黄土力学特性的影响；并以改性黄土的室内试验结果作为数值计算模型的依据，对季冻区黄土路基进行数值计算。主要研究工作和结论如下：

（1）通过无侧限抗压试验，分别探究水泥掺量、纤维长度、纤维掺量以及冻融循环次数对改性黄土抗压强度的影响；运用灰色关联度理论分析不同冻融循环次数下各因子的关联度大小；利用Lasso回归分析建立改性黄土无侧限抗压强度回归模型。前5次冻融循环内，水泥的掺入使改性黄土强度损伤较小，冻融循环15次后改性黄土抗压强度趋于稳定。

（2）通过三轴压缩试验，分析素黄土和改性黄土在冻融循环作用下的应力应变特性、黏聚力、内摩擦角等的变化规律；引入冻融损伤系数，推导改性黄土经多次冻融循环的黏聚力与未经冻融循环黏聚力之间的拟合关系式；与素黄土相比，相同冻融循环次数下，改性黄土的黏聚力和内摩擦角均有所提高。利用电镜扫描测试，对比改性黄土微观形貌，探讨其在不同围压下的断面微观形态。

（3）以室内相关试验结果为依据，以季冻区黄土路基力学参数进行数值计算。在10年的冻融循环内，温度对黄土路基的作用范围约为2 m；第10年冻融循环融化期结束时路基表面无冰，但在黄土路基下方大约1.5 m处水仍旧以冰的形式存在；在第10年冻融循环结束时，路基顶面3 m以下土体应力没有明显变化，路肩2.5 m以下土体应力没有明显变化。

Loess is widely distributed in northwest China, along with the seasonal freezing and thawing, the mechanical properties of loess change, and it is very easy to cause engineering problems, such as subgrade frost swelling, thawing and sinking, mud overturning, etc. In this paper, we use cement and polypropylene fiber to modify loess in northwest China, and study the effects of cement mixture, fiber length, fiber mixture and the number of freeze-thaw cycles on the mechanical properties of reinforced loess; and determine the mechanical parameters based on the indoor test results of reinforced loess, and make numerical calculations on the loess subgrade in the seasonal freezing area. The main research works and conclusions are as follows:

(1) The effects of cement admixture, fiber length, fiber admixture and the number of freeze-thaw cycles on the compressive strength of reinforced loess were investigated through the unconfined compressive strength test; the correlation magnitude of each factor under different number of freeze-thaw cycles was analyzed by using gray correlation theory; the regression model of unconfined compressive strength of reinforced loess was established by using Lasso regression analysis. Within the first five freeze-thaw cycles, the cement incorporation caused less damage to the strength of reinforced loess, and the compressive strength of reinforced loess stabilized after 15 freeze-thaw cycles.

(2) Through triaxial compression tests, the changes of stress-strain curves, cohesion and internal friction angle of plain loess and reinforced loess under freeze-thaw conditions were analyzed; freeze-thaw damage coefficients were introduced to derive the fitted relationship between the cohesion of reinforced loess after multiple freeze-thaw cycles and the cohesion without freeze-thaw cycles. Compared with the plain loess, the cohesion and internal friction angle of the reinforced loess were improved under the same number of freeze-thaw cycles. The damage patterns of the reinforced loess under different circumferential pressures were explored by comparing the microscopic morphology maps using electron microscope scanning tests.

(3) Based on the results of the relevant indoor tests, the mechanical parameters of the loess subgrade in the seasonal freezing zone were set for numerical calculation. Within the 10-year freeze-thaw cycle action, the temperature action on the loess subgrade is about 2 m. There is no ice on the subgrade surface at the end of the thawing period of the 10th year freeze-thaw cycle, but water still exists in the form of ice at about 1.5 m below the loess subgrade. At the end of the 10th year freeze-thaw cycle, there is no significant change in soil stress below 3 m on the top surface of the subgrade and no significant change in soil stress below 2.5 m on the shoulder.