Shale displays obvious anisotropy due to the orientation of the rock mineral particles during the deposition process and its layered structure of weak surfaces. Potassium chloride can inhibit the hydration expansion of shale and improve its strength, although wellbore instability is extremely common in layered shale during drilling. In this paper, core samples were collected at different bedding angles (0°, 30°, 60° and 90°) to study the mechanical and anisotropic characteristics of shale saturated in different concentrations of potassium chloride. The laboratory uniaxial compressive strength show that the failure mode of the shale is related to the angle of the bedding plane, and the failure always induces tensile stress at 0° and 90°. At 30° and 60°, shear sliding occurs along the bedding plane. With increasing bedding angle, the strength first decreases and then increases, so the anisotropic strength characteristic curve is approximately U-shaped. The inhibition of clay swelling is optimum at a 4% KCl concentration. A uniaxial compressive strength prediction model is established by an artificial neural network, and the influencing factors are analyzed. The calculation of uniaxial compressive strength is improved, and the problem of predicting uniaxial compressive strength is solved. This study not only elucidates the mechanical anisotropy of shale but also provides basic data on borehole stability and hydraulic fracturing.

页岩显示出明显的各向异性，这是由于沉积过程中岩石矿物颗粒的取向及其弱表面的分层结构所致。氯化钾可以抑制页岩的水合膨胀并提高其强度，尽管钻井过程中层状页岩的井眼失稳非常普遍。在本文中，以不同的层理角度（0°，30°，60°和90°）收集岩心样品，以研究在不同浓度的氯化钾中饱和的页岩的力学和各向异性特征。实验室单轴抗压强度表明，页岩的破坏模式与层理平面的角度有关，并且破坏总是在0°和90°引起张应力。在30°和60°时，剪切会沿着层理平面发生滑动。随着床角的增加，强度先减小然后增大，因此各向异性强度特性曲线近似为U形。KCl浓度为4％时，对粘土膨胀的抑制作用最佳。利用人工神经网络建立了单轴抗压强度预测模型，并对其影响因素进行了分析。改进了单轴抗压强度的计算，解决了预测单轴抗压强度的问题。这项研究不仅阐明了页岩的机械各向异性，而且还提供了有关井眼稳定性和水力压裂的基础数据。利用人工神经网络建立了单轴抗压强度预测模型，并对其影响因素进行了分析。改进了单轴抗压强度的计算，解决了预测单轴抗压强度的问题。这项研究不仅阐明了页岩的力学各向异性，而且还提供了有关井眼稳定性和水力压裂的基础数据。利用人工神经网络建立了单轴抗压强度预测模型，并对其影响因素进行了分析。改进了单轴抗压强度的计算，解决了预测单轴抗压强度的问题。这项研究不仅阐明了页岩的机械各向异性，而且还提供了有关井眼稳定性和水力压裂的基础数据。