大样本岩土参数概率分布的正态信息扩散推断

Assessment of probability distribution of large samples of geotechnical parameters by using normal information spread estimation method

在重要的岩土工程中,确定大样本岩土参数的概率分布对岩土工程稳定性和可靠性分析有着极其重要的意义。为此,基于积分均方误差最小计算得到的最优窗宽,提出了推断大样本岩土参数概率密度函数的正态信息扩散法。该方法基于信息扩散原理,从试验样本和信息论的角度出发,充分利用样本提供的数据信息,而不是先假定成经典的概率分布曲线拟合检验,数学意义和物理意义更加充分和严密。以推断压缩指数Cc的概率密度函数为例,分析了窗宽对信息扩散估计结果的影响规律,说明了该方法在大样本岩土参数概率密度函数推断方面的合理性。用该方法推断了大样本岩土抗剪强度参数的概率密度函数,通过K-S检验,验证了该方法的正确性和实用性。

It is much more significant for assessment of probability distribution of large samples of geotechnical parameters to study the stability and reliability of geotechnical engineering in some important geotechnical engineerings. Thus normal information spread estimation method （NISEM） is proposed to generate probability density functions of geotechnical parameters with large samples based on the optimal bandwidth which is received by the minimum mean integrated square error. The method proposed in this paper based on the information spread principle, instead of assuming the fitting test of classical probability distribution curves, makes full use of the data sample provided from the view of the test sample and the information theory. It presents the mathematical and physical performances more fully and tight. The influence of bandwidth on the results of information diffusion estimation has been studied by the inference of probability density function of compression index （Cc）. The results show the reasonability of NISEM, which is used to infer probability density function of geotechnical parameters with large samples. At last, probability density functions of rock and soil shear strength parameters have been inferred by using NISEM. By K-S test, the correctness and practicability of NISEM are verified.