地质雷达超前预测不良地质体图像的智能识别

Intelligent identification of unfavorable geologic body by GPR image

摘要: 采用GPR地质雷达进行隧道超前地质预报,当掌子面前方存在溶洞时,地质雷达时间剖面常出现双曲线特征图形。针对这一比较成熟的先验知识,提出采用神经网络模式识别方法自动识别地质雷达图像中的双曲线,以此实现智能预测掌子面前方溶洞这一不良地质体。结果表明,这一方法是可行的。

Abstract: In the field of tunnel fore geological forecast,when karst caves exist in front of work face,the time sec-tion of GPR usually has hyperbola character.The paper,according to the mature knowledge,put forward a method using nerve network mode to identify hyperbola in GPR image and to realize intelligent forecasting of karst caves in the front of work face,the results indicate the method is feasible.