Development of an AI-Based YOLOv8 System for Microcrack Detection in Aircraft Jet Engine Components using Borescope Images

Abstract

The timely detection of microcracks in jet engines is critical for aircraft safety, yet manual inspections are prone to human error, and existing AI models face limitations in accuracy and reliability. This study addresses these challenges by developing a novel deep learning framework for the detection and classification of microcracks in aircraft jet engines using the YOLO v8 algorithm. Through rigorous development and testing, the research demonstrates significant advancements over traditional computer vision techniques and previous deep learning approaches. The research utilized dataset of 27,708 high-quality borescope inspection images collected from three major Nigerian airports. The model achieved perfect precision (1.000), 87.5% recall, 100% specificity, 90.9% accuracy, and an F1 score of 0.933. The mean Average Precision reached 98.9% at IoU threshold 0.5 and 95.2% across the IoU range 0.5:0.95, confirming exceptional detection performance. Training and validation metrics showed effective model convergence without overfitting, while confusion matrix analysis revealed robust classification capabilities with no false positives. The system successfully detects various crack morphologies across different lighting conditions and surface textures, as demonstrated in visual results. A dedicated Windows application provides a practical interface for maintenance technicians to integrate this technology into existing workflows. This research directly addresses aviation safety challenges by enabling earlier and more reliable detection of potential failures in jet engine components, such as, compressor blades, fan blades, compressor casing, turbine blades, turbine disks, turbine vanes/nozzles, bearings and bearing housing, nozzle guide vanes, exhaust cone and ducting thereby enhancing maintenance efficiency and reducing operational risks. The methodology established provides a foundation for developing comprehensive AI-assisted inspection tools for the aviation industry.