Artificial intelligence-assisted method of structural health monitoring of subway shield tunnel based on insulation degradation location

Abstract

Stray current leakages inevitably arise to seriously threaten the reliability of the subway tunnel structure and third-party system through electrochemical corrosion. Location of stray current leakage provides novel approach to mitigate stray current corrosion and valid guidance for maintenance of subway tunnel. The present study is dedicated to develop an integrated network-based method for locating the insulation degradation in subway tunnel. In this work, a predictive model based on LWQPSO-SOM algorithm is designed to cluster the risk level of stray current leakage through data mining. Then, an evaluation index Phigh is proposed to calculate the probability of high-risk index in the clustering results. Identification results of leakage zone of stray current is effectively validated through the measurement of rail-to-earth conductance. Results showed that the distribution of Phigh is highly related with rail-to-earth conductance distribution, indicating that the proposed method is applicable for stray current leakage location in the subway tunnel and potentially applicable in engineering fields.