图  1  钻孔中应力观测示意图

地壳应力不均质性（深绿色箭头表示应力方向）；地壳岩体结构面如裂隙、断层等（红线）

Figure  1.  Illustrations of in-situ stress observations in boreholes

The crustal stress heterogeneity is illustrated by the orientations of the green arrows, and crustal rock mass discontinuities such as fractures and faults are illustrated by the red lines.

图  2  利用深度学习自动拾取钻孔成像中的崩落示例

超声波成像测井的振幅、走时数据，通过深度学习手段自动拾取崩落等地应力指示物信息

Figure  2.  Example of automatic breakout identification from borehole image logs via deep learning

Amplitude and travel–time data of ultrasonic imaging logging are used to automatically extract the information of in-situ stress indicators such as borehole breakouts by means of deep learning.

图  3  瑞士Bedretto地下实验室应力表征的概况

图中X表示正东方向，Y为正北方向a—钻孔成像展示的一处钻孔崩落；b— 钻孔成像及其反演钻孔微小椭圆度变形

Figure  3.  Overview of the in-situ stress characterization in the Bedretto Underground Laboratory, Switzerland

(a) Breakout identified on the borehole image log; (b) Borehole ellipticityX represents the east direction, and Y represents the north direction.

图  4  利用钻孔中天然裂隙导水性开展地壳应力场和岩体摩擦系数反演的示意图

将钻孔成像和温度测井识别并分类的导水/不导水裂隙，作为输入数据进行应力场和地壳岩体摩擦系数的反演，获得联系裂隙导水性和剪切应力临界性的最大似然概率

Figure  4.  Illustration of the inversion of the crustal stress and rock mass frictional coefficient via natural fracture hydraulic conductivity

Natural fractures are identified and classified through borehole image logs and temperature logs; they serve as input for the inversion of in-situ stress and the crustal rock mass frictional coefficient, which attains maximum likelihood linking fracture hydraulic conductivity and shear stress criticality.