A novel framework for void detection and health evaluation of cold-region diversion tunnels using GPR and multiattribute modeling

Voids in diversion tunnel linings can induce cracking and leakage, compromising structural integrity. To assess the health condition of a cold-region diversion tunnel lining in China, ground penetrating radar (GPR) was employed for void detection, supplemented by statistical data on voids, cracks, and leakage. A multi-attribute evaluation model was developed by integrating dynamic, subjective, and objective indicator weights using game theory, implemented via improved cloud-evidence theory. The model’s effectiveness was validated against alternative methods. GPR identified 127 lining voids, with drilling verification confirming high reliability; however, detecting ice-filled voids remained challenging. Evaluation results indicate the lining is in a slightly deteriorated condition, adversely affecting operational safety. The proposed improved cloud-evidence theory method effectively addresses randomness, ambiguity, and conflicting evidence in indicator assessment. This study provides substantial guidance for maintenance decision-making in cold-region diversion tunnels.