Monitoring of Sleeve Grouting Process based on Enhanced YOLOv4 Algorithm with African Vultures Optimization

The stability and structural integrity of prefabricated construction components are reliant on the quality of grouting in sleeve connections. Since the sleeves are submerged in the structure, there is yet no reliable Nondestructive Testing (NDT) technique for evaluating the grouting quality in such intricate configurations. While earlier research has shown that NDT methods are used to single-socket grouting connections, more complex sleeve configurations still provide a problem. So, an advanced method named Enhanced You Only Look Once version 4 (EYOLOv4) is employed for monitoring the grouting process in sleeve connections. In order to fine-tune the YOLOv4 algorithm's hyperparameters for better performance in sleeve grouting process monitoring an African Vultures Optimization (AVO) approach is proposed. This research hybridized the EYOLOv4 and AVO (HEYOLOV4-AVO) for monitoring sleeve grouting process. In comparison with conventional methods such as Piezoelectric Ceramic Scheme (PCS), and Defect-Detectable and Repairable Half-Grouted Sleeve (DDRHGS), the proposed HEYOLOV4-AVO obtains greater health rate of 5.89.