人工神经网络方法在海州矿边坡不稳定区圈定中的应用

APPLICATION OF ARTIFICIAL NEURAL NETWORK METHOD INTO DETERMINATION OF INSTABILITY ZONE OF SLOPE IN HAIZHOU MINE

边坡稳定性受工程地质、水文地质条件和人工活动(开挖或防护)等因素的影响,这些因素决定了滑坡的具体形态。然而,上述因素与滑坡形态之间很难找到某种确切的关系。海州矿非工作帮边坡从建矿到目前,已发生过70多次滑坡,规模较大且有详细资料记录的滑坡60多次。将这些丰富的滑坡资料进行整理,并尝试运用人工神经网络方法,将工程地质因素,如滑落层的内摩擦角和傾角、滑落层走向与边坡走向间的夹角、剪切带长度、降雨量、断层、坡高等因素作为网络的输入,以滑坡体的长度、宽度和顶部裂缝距坡顶的距离作为网络的输出。各因素与滑坡体的长度、宽度和顶部裂缝距坡顶的距离之间的非线性动力学关系则由网络的结构描述。网络训练好以后便可用于对潜在不稳定区范围的预报和圈定。这样,一方面可以有针对性地采取整治措施;另一方面,可以据此进一步分析和验证按传统方法所圈定的边坡是否存在滑坡的危险性。

The stability of slope is affected by many factors,which determine the specific shape,such as engineering geology,hydrogeologic condition and the activity of people (excavation or guard). However,it is difficult to find the definitive relationship among those factors and the specific shape. More than seventy times of landslides have happened in the unworking band of Haizhou mine since its foundation,and there more than sixty times of largescale ones were recorded in detail. Based on the abundant landslide data,the method of artificial neural network is used to describe the nonlinear relationship mentioned above. In the method,the input parameters include the friction angle and dip angle of the slide layer,the separation angle between the slide surface and slope trend,the length of the shear zone,rainfall,fault and the hight of the slope,while the output parameters include the length and width of gliding mass,and the distance between the crack and tip of the gliding mass. The result calculated by the network can be used to estimate the shape of other potential gliding mass only after the training of the network is completed. Then some effective measures can be taken to deal with the landslide,and also,the presented result can be used to validate the instability zone got in traditional ways.