LAUSR

Abstract Karst is a geological phenomena in which water mainly dissolves soluble rocks, supplemented by mechanical actions, such as water erosion, potential erosion and collapse. During the design and construction of submarine tunnel, the fine detection of karst cave shape plays an important role, which is related to the construction process of karst area and the stability evaluation in the later stage. However, due to the irregularity and complexity of submarine karst morphology, it is difficult for the existing detection technology to realize the fine detection of cave shape. In this paper, a method of submarine karst morphology detection based on multi-frequency ultrasound is proposed on the basis of laying the ultrasonic probe down the borehole to the rock detection area by cable. Firstly, according to the morphological characteristics of submarine karst, the acoustic reflection model of submarine karst surface is established. Combining with the mechanism of ultrasonic acoustic propagation, the principle of automatic frequency matching is established to ensure that the detection system is always in the best working state, so as to obtain more effective and accurate acoustic detection data. Then, the principle of synthetic aperture imaging is used to realize the various submarine karst caverns. High-precision image of deep horizontal section is presented, and an improved image segmentation method is proposed to extract the contour features of horizontal section at different depths of submarine karst. Search criteria for retrieving calibration value of contour line size are established, and a precise determination system of ultrasonic scanning acoustic beam ranging value is formed to realize the radial distance inversion of the contour of submarine karst horizontal section. Finally, the wheel of submarine karst cavity section is completed. On the basis of fine profile measurement, the depth information of submarine karst is superimposed, and the three-dimensional shape reconstruction of submarine karst is realized by synchronous triangular network method. The feasibility and accuracy of the method are verified by field verification and result analysis.