Urban Road Underground Defects Detection Using the Resistivity Method with an Irregular Array

Abstract While the resistivity method offers clear advantages for detecting underground defects beneath urban roads, the complex surface conditions of these roads preclude the deployment of regular arrays. This complexity significantly challenges conventional resistivity detection methods for underground defects. This study designed a high-density resistivity detection observation system adaptable to the deployment of irregular arrays along urban roads. This method obtains high-dimensional apparent resistivity data and employs Principal Component Analysis (PCA) for its dimensionality reduction. It introduces an improved position update method within the Chimpanzee Optimization Algorithm (ChOA), combined with Extreme Learning Machine (ELM), to propose a PCA-ChOA-ELM combination algorithm for inversion. This is then compared with BP (Back Propagation Neural Network), GABP (Genetic Algorithm optimized Back Propagation Neural Network), and ELM. To further test the PCA-ChOA-ELM resistivity inversion’s effectiveness in detecting underground defects beneath urban roads with an irregular array, a geological-geophysical model of underground defects was established. Numerical and physical experiments were combined to perform inversion imaging on models of underground defects, both with and without water. The research results show that the high-density resistivity method based on PCA-ChOA-ELM with an irregular array can adapt to the detection of underground defects under urban roads, can better reflect the electrical characteristics, location, and distribution of underground defects under urban roads, and achieve efficient and precise positioning of underground defects under urban roads.