Active frequency selective surface (AFSS) is composed of the traditional frequency selective surface (FSS) unit patterns and a series of impedance elements loaded between them, performing excellent adjustable electromagnetic characteristics.The quality of solder joints seriously affects the frequency response characteristics of AFSS, while a series of inspection methods such as manual visual inspection unable to meet the increasingly stringent inspection standards. Searching for efficient and accurate means of detecting soldering defects has become the focus of attention of many researchers. Taking AFSS based on array capacitor as an example, a solder joint defect detection method based on active infrared thermal imaging and improved YOLOv3 is proposed. The active infrared detection system for solder joint defects is built, and the thermal infrared image data set of capacitor solder joint defects is established through data enhancement. Aiming at the feature that the defects in the data set are concentrated on small scales, the feature loss is reduced by introducing the SPDConv module in the backbone network, and the improved CBAM attention module is introduced in the residual module to focus on the small target features. At the same time, the RFB-Conv module is built for replacing the CBL stacked blocks in the original neck structure of YOLOv3, which helps to homogenize the delicate sensory field. The experimental results show that the method has excellent performance in the detection of AFSS solder joint defects based on array capacitors, which provides a good idea for the detection of AFSS.
发表评论