基于Darknet网络和YOLO4的实时电路板故障检测算法

Real-time PCB Fault Detection Algorithm Based on Darknet Network and YOLO4 Algorithm

针对现有的接触式电路板故障检测方法难以应用到大规模集成电路故障检测中的问题,提出一种实时、非接触式的基于深度学习的电路板故障诊断算法;建立PCB板缺陷检测和元器件识别图像数据集,并采用数据增强技术,对数据进行数据增强来提高训练的数据量,以提升模型检测精度和鲁棒性;基于Darknet框架和YOLO4算法训练得到元器件检测模型,并通过采用k-means聚类算法设计合理的Anchors,使得模型具备多尺度缺陷检测的功能;使用图像配准算法在红外图像和可见光图像上实现配准和融合;根据PCB板设计时划分的功能区域,利用测温热像仪连续采集5个该区域的平均温度,通过判断5个平均温度之间的关系从而判断短路或者短路状态;经过试验测试,使用预先设置好故障的电路板作为实验对象,通过采集实验对象运行过程中的红外和可见光图像数据,基于设计的故障检测模型,不仅能够实时且有效地识别出元器件位置,并能够直观地标识出现短路、短路故障元器件;经过实际应用,能够满足设备运行时的实时电路板故障检测工程应用。

Aimed at the problem that existing contact circuit board fault detection methods are difficult to be applied in large scale integrated circuit fault detections,a real time contactless circuit board fault diagnosis algorithm based on deep learning is proposed.The image data set of PCB board defect detection and component recognition is established,and the data enhancement technology is used to enhance the data volume of training,and improve the accuracy and robustness of the model detection.The component detection model based on Darknet framework and YOLO4 algorithm training is obtained,and K-means clustering algorithm is used to make the model have the multi-scale defect detection function.Image registration algorithms are used to register and fuse infrared and visible images.According to the divided functional area of the PCB board design,the average temperature of five areas is collected continuously by the thermometry thermal imager,and the short circuit or short circuit status is judged by the relationship between five average temperatures.After testing,the pre-set faulty circuit board is used as the experimental object,the infrared and visible image data is collected during the operation of the experimental object,based on the designed fault detection model,the real-time and effective identification of short-circuit faults with component and regional components is realized.and it can meet a engineering application of real-time circuit board fault detection for equipment running.