基于深度学习的卷包设备清洁保养质量判别系统设计

Deep learning-based discriminant system design of cleaning and maintenance quality of cigarette maker and packer

为解决烟草行业设备保养人工检查效率低、标准不一等问题,设计了一种基于深度学习的卷包设备清洁保养质量判别系统。系统主要包括三个模型,基于深度学习的保养部位识别模型,判别采集到的图像是否是正确的保养部位;基于深度学习的脏物检测模型,从采集到的保养图像中检测出不合格的脏物;融入工艺知识的保养质量判别模块,根据检测到的脏物和保养部位信息,判别保养是否合格。以武汉卷烟厂卷包车间为例进行现场测试,结果表明:系统对卷包设备的清洁保养质量判别准确率达到86.3%,满足实际生产中对卷包设备清洁保养质量的自动化判别需求,具备良好的泛化性能。

Aiming at low efficiency and non-identical standards of manual inspection of equipment maintenance in the tobacco industry,a deep learning-based system for judging the maintenance quality of cigarette maker and packer was designed.The system mainly consists of three models,including the maintenance part recognition model based on deep learning to determine whether the collected image is the correct maintenance part,the dirt detection model based on deep learning to detect unqualified goods from the collected maintenance images,and the maintenance quality judgment module integrated with process knowledge to judge whether the maintenance is qualified according to the detected goods and maintenance part information.On-site test was conducted by taking the Wuhan Cigarette Factory’s wrap-up workshop as an example.The results show that the system’s accuracy rate of judging maintenance quality of cigarette maker and packer reached 86.3%,which meets the demand for automatic identification of maintenance quality of cigarette maker and packer in actual production and has good generalization performance.