基于深度学习的风叶螺母检测

The Detection of Fan Blade Nuts Based on Deep Learning

风叶螺母装配质量主要依靠目视、打点进行检查,完全依赖员工操作,基于生产线节拍、重复动作、视觉疲劳等客观条件,存在螺母漏打导致的风叶脱落、噪音异响等关键装配质量异常风险。针对该问题,本项目提出了一种基于CNN算法的风叶螺母在线视觉检测方法,通过运用模糊识别算法智能过滤非检测目标,该方法具备模板学习能力,还可以实现自动扫描、自动关联产品信息以及质量数据分析和追溯功能。

The assembly quality of fan blade nuts mainly relies on visual inspection and dot inspection,and completely depends on employee operation.Based on objective conditions such as production line beats,repetitive movements,and visual fatigue,there are abnormal risks of key assembly quality such as fan blade falling off and abnormal noise caused by nut leakage.To solve this problem,this project proposed an online visual inspection method for fan blade nuts based on CNN algorithm.By using fuzzy recognition algorithm to intelligently filter non-detected targets,this method has template learning ability,and can also realize automatic scanning,automatic association of product information,quality data analysis and traceability functions.