岩石参数Bayes估计中验前样本可信度的研究

针对岩石试验属破坏性试验，不易取得大量样本数据这一特点，提出在利用验前信息对岩石力学参数进行Bayes估计时，应首先考虑验前信息的可信度。通过统计学假设检验中二类错误理论推导了正态分布下验前信息与验后信息的相容性和可信度的公式，并结合康家湾矿的岩石单轴抗压强度数据进行计算。结果表明，运用该公式计算验前信息可信度是可行的，为利用验前信息进行参数估计提供了理论依据。

Credibility of sample's prior information in estimation of rock mechanics parameters by Bayes method

Since the mechanical test on rock is a kind of destructive test, it is difficult to derive massive sample data for determining mechanics parameters. The credibility of the prior information should be considered at first in estimating rock mechanics parameters by the Bayes method. The formulas for calculating the compatibility and credibility of the pre-information and post-information under the condition of normal distribution are developed based on the assumption of second kind-mistake theory of statistics. This result provides the theoretical basis of utilizing the prior information to carry out the estimation of parameters. The proposed method is applied to calculate the data of uniaxial compressive test for the rock in a mining project. The result shows that the formula for calculating the credibility of prior information is feasible.