Compact Muon Solenoid Detector Quality Assurance Using Machine Learning

Background Precision and accuracy in quality assurance are critical for high-energy physics experiments. The Compact Muon Solenoid experiment at the Large Hadron Collider is undergoing an upgrade with the High-Granularity Calorimeter, requiring the inspection of numerous silicon detectors and hexaboards. Manual inspection methods, such as Coordinate Measuring Machines, are time-consuming and labor-intensive. This study investigates an automated quality assurance framework using flatbed scanners and deep learning algorithms. Methods The EPSON v850 flatbed scanner was selected for its high resolution and large scan area. Three approaches were explored for extracting stepped holes from hexaboard scans: Circular Hough Transform (CHT), printed circuit board edge detection, and bottom-most hole detection with CHT. Signal pad detection was performed using the YOLOv8 architecture, while concentricity offsets were measured with CHT. The pixel-to-real-world coordinate mapping was calibrated using dots per inch measurements, and the framework was optimized to reduce processing time. Results The automated inspection framework successfully localized 92 stepped holes per hexaboard and detected signal pads with high accuracy. The concentricity offset was calculated with a precision of 303 microns. The optimized system reduced processing time to approximately 2 m 6 s ± 692 ms at 3,200 dots per inch, compared to several minutes with manual methods. However, noise from dust and scanner skewness introduced occasional inaccuracies. Conclusions The proposed framework offers a scalable and efficient solution for quality assurance in the Compact Muon Solenoid experiment's High-Granularity Calorimeter upgrade. By leveraging deep learning and advanced image processing techniques, the system significantly enhances inspection speed and accuracy. Future improvements, including noise mitigation and enhanced scanner hardware, could further optimize performance. This methodology holds promise for broader industrial applications requiring high-precision inspections.