寒区渠基黏土热参数最优概率分布

Optimal Probability Distribution of Thermal Conductivity Parameters of Canal Foundation Clay in Cold Regions

为分析寒区渠基黏土热参数的随机分布特征及概率分布模型,以寒区渠基黏土的导热系数为样本,结合经典分布拟合法、多项式逼近法、最大熵法和正态信息扩散法,分别对寒区渠基黏土热参数的概率分布规律进行了研究。首先通过分析热参数的离散性,并比较概率分布曲线、拟合检验值和累计概率分布值,对不同方法描述热参数随机性的优劣进行了评价;然后,基于寒区渠基黏土热学参数对温度的敏感性,提出了一个可以达到理想拟合精度的寒区渠基黏土热参数概率推断的区间取值标准。研究结果表明:寒区渠基黏土的热参数具有随机变量的特征;正态信息扩散法可以描述热参数样本的随机波动性;在4种方法中,正态信息扩散法的拟合精度最高。使用3.5σ法,将[μ-3.5σ,μ+3.5σ](μ为随机变量的均值,σ为标准差)作为概率函数推断时的取值区间,同时考虑偏度的影响,可使得累计概率值达到1.000 0的精度,能够较准确地推断热参数的概率分布函数。

In order to analyze the random distribution characteristics and probability distribution model, taking the thermal conductivity of canal foundation clay in cold regions as a sample, the probability distribution laws of the thermal parameters of frozen soils were inferred by using the classical distribution fitting method, polynomial method, maximum entropy method, and normal information diffusion method. Through analyzing the dispersion of thermal parameters and comparing the probability distribution curve, fitting test values, and cumulative probability distribution values, the advantages and disadvantages of different methods to describe the randomness of thermal parameters were evaluated. The results show that the thermal parameters of canal foundation clay are variable in cold regions. The random fluctuation of thermal parameters can be described by the normal information diffusion method. Among the four methods, the fitting precision of the normal information diffusion method is the highest. Based on the sensitivity of the thermal parameters of canal foundation clay to temperature, a standard with an ideal fitting precision is put forward for determining the probabilistic inference interval of the thermal parameters of frozen soils.