Data Processing Approaches for Electromagnetic Survey and Inspection of Subsurface Structures

Modern infrastructure management faces significant challenges in detecting and inspecting subsurface utility objects and structures, such as piping networks, electrical and cable systems, and other underground construction elements. For effective infrastructure maintenance, surveying and inspection are critically important and require specialized equipment to locate and identify these objects. However, interpreting signals from these devices remains complex.

This study focuses on developing data signal processing techniques for subsurface exploration and inspection equipment by applying machine learning algorithms and statistical data analysis to improve the accuracy of object identification and localization, as well as to detect underground electrical leakage with high precision. The proposed technique integrates data from three types of equipment based on electromagnetics principle, which are pipe and cable locators (PCL), ground-penetrating radar (GPR), and stray current detectors (STD). Significant features obtained from each device were extracted and analyzed using dedicated algorithms and methods. The analysis results demonstrated an identification accuracy of over 80% for subsurface objects within 1 meter across various mediums, including plastic, metal, and concrete. The integration of multi-source data processing leveraged the strengths of each device, enhancing the reliability of analysis and reducing errors associated with relying on a single data source. Ultimately, the developed system shows strong potential for application in infrastructure planning, design, and maintenance, while also reducing the risk of unintentional excavation damage to subsurface structures.