基于YOLO3的活塞连杆智能防错系统开发

The Engine Piston Rod Assembly AI Error Proofing System Based on YOLO3

活塞连杆及连杆盖需按要求配对装配,如出现混装,会导致发动机异常磨损、异响、抱死等问题,严重影响客户体验。目前活塞连杆及连杆盖装配过程质量靠操作工目视检查,无有效防错,属于过程监控盲区。为解决该问题,基于YOLO3开发智能视觉系统监控整个连杆装配过程,使用Python编程,并以NVIDIA Jetson nano为终端平台,通过不断优化计算模型,最终实现了低成本的活塞连杆装配全过程监控。该视觉系统完全兼容工厂各种变化的环境,无需单独补光,在不增加操作节拍、无需人员额外操作的基础上做实现高效防止活塞连杆混装问题,且准确率达到99.5%以上。

The connection rod assembly is particularly important to the engine's quality.The rod and the caps shall be assembled in pairs.If not,it could cause engine abnormal worn noise or engine broken which got customer walk home.To solve the issue,this paper develops an AI vision error proofing system based on the YOLO3 via Python and integrated in the NVIDIA Jetson nano platform.The AI vision system can adapt changed environment and get over 99.5%right identification without additional light or operation time.