Maintenance of underground pipelines and sewage systems is a significant challenge since they are underground. A sewer pipe surrounded by voids can lead to groundwater entering the voids, thereby losing the support of the surrounding soil. Groundwater infiltration leads to further structural damage resulting in a collapse of the ground surface over the pipe.
This research focuses on using in-pipe Ground Penetrating Radar (GPR) to detect the voids composed of air or water. gprMax software is used to simulate the radargrams containing a different composition of voids around the sewer pipeline. Following this, three different Machine-learning models are used to identify the void characteristics and classify them based on material. The YOLOV3 method gives the most accurate results in several scenarios that were tested.
In addition to this, image processing methods were also used to extract the hyperbolic features to check for further improvement. It was concluded that for cases where different layers of sand with differing permittivity values are present around the pipe, feature extraction added with YOLOV3 gave accurate results. Hence, based on simulations and comparisons of the three methods, YOLOV3 was more suitable for void detection and had a detection accuracy of 99%. Currently, the GPR survey is performed manually.
Therefore, a symbiotic relationship between the operator and the algorithm can be created by applying machine learning to current in-pipe GPR practices.
To join the presentation via Microsoft Teams click here