一种智能化基站工艺验收框架研究

Research on an intelligentized pipeline for the acceptance of base station

当前的基站工艺验收工作流程主要是以人工操作为主,IT手段为辅,整个过程流程不仅繁琐,而且效率极其低下。基于基站工艺验收工作中收集到的海量图像数据,本文设计了一种智能化的验收框架,开发手机APP帮助装维人员采集并上传拍摄的验收图像,利用先进的物体检测、机器学习等AI模型构建出原子服务能力,以自动化的方式进行图像的查重、防翻拍、检测和识别,同时打造出标准的图像标注和难样本收集流程,方便后续验收能力的扩展和模型精度的提升。

The work of base station acceptance involved much manual operation,which leaded to not only the standard of acceptance was inconsistent,but the work effi ciency was pretty low because of the high rework rate resulted from cheating behavior.For properly utilizing a great deal of image collected from base station acceptance work,an intelligentized pipeline was proposed in the help of IT and AI technology.Firstly,a mobile phone APP was developed to promote the effi ciency of image collection work.Secondly,image check algorithm and antiremake algorithm was designed to prevent cheating behavior.Third,AI acceptance ability was developed based on the state-of-the-art object detection and character recognition model,and deployed in the central cloud platform.Additionally,a hard sample collecting fl ow was presented for helping to extend or regularly improve AI acceptance ability.The effi ciency of base station acceptance work could be improved by proposed pipeline.