In the process of tunnel excavation, it is necessary to understand the geological situation in front of the palm in real time, and the methods of forward prediction of tunnel construction commonly used at this stage include seismic method and geological radar method, which are often used jointly by two methods in a large number of engineering cases, and the results of the two methods are verified by each other.

In the engineering practice of geological radar advanced forecasting, the general workflow is to first use geological radar to collect information in front of the palm of the hand, form a geological radar signal profile, after collection, copy the data from the acquisition instrument to the post-processing system, and then by professionals to process and judge each section. This series of work takes more time and labor and affects the construction cycle.

      In response to this problem, we use machine learning methods to develop a geological signal automatic identification system, and embedded in the geological radar host. After the data acquisition is completed, it can quickly analyze and identify the collected data, filter out the detection data that may be problematic, shorten the workflow, improve production efficiency, reduce production costs.