The Hoek-Brown (H-B) failure criterion is an empirical failure criterion. The estimates of Hoek-Brown criterion parameters, such as the geological strength index (GSI) and the disturbance coefficient (D), are usually subjective. This paper focused on modifying the initial estimates of GSI and D to improve the accuracy of parameters. The nonlinear regression model (NLRM) of the Hoek-Brown failure criterion was proposed to analyze the rock parameters by using the sensitivity analysis and the displacement equation of the surrounding rocks. Then, a reasonable back analysis method was developed by introducing the differential evolution (DE), which was used to accurately obtain the parameters of the Hoek-Brown failure criterion in practical engineering. This method was successfully used to analyze the stability of the roadway in a deep coal mine. The results showed that the NLRM can better reflect the relationship between GSI, D, μ and the displacement of roadways, and the back analysis results are consistent with the filed monitoring results. This method can provide a helpful reference for modifying the influence of empirical and subjective factors on H-B parameters selection, and improving the accuracy of Hoek-Brown criterion parameters in the similar engineering applications.

Hoek-Brown (HB) 失效准则是一种经验失效准则。Hoek-Brown 准则参数的估计，例如地质强度指数 ( GSI ) 和扰动系数 ( D )，通常是主观的。本文重点修改GSI和D的初始估计以提高参数的准确性。提出了Hoek-Brown破坏准则的非线性回归模型(NLRM)，利用敏感性分析和围岩位移方程对岩石参数进行分析。然后，通过引入微分演化（DE）发展了合理的反分析方法，用于在实际工程中准确获取Hoek-Brown破坏准则的参数。该方法已成功应用于某深部煤矿巷道稳定性分析。结果表明，NLRM能够较好地反映GSI、D、μ之间的关系和巷道位移，反分析结果与现场监测结果一致。该方法可为修正经验和主观因素对HB参数选择的影响，提高Hoek-Brown准则参数在类似工程应用中的准确性提供有益参考。