基于数字孪生的选煤厂厂区人员视频定位方法

Video location method of production personnel in coal preparation plant based on digital twin

为提升生产人员安全视频实时定位效果,针对轮廓边缘清晰度低、安全区域与生产人员轮廓难以分割的问题,研究基于三维数字孪生的选煤厂厂区生产人员安全视频实时定位方法。数据层通过调用物理层内各类别传感器,采集选煤厂区实时视频数据,清洗与增值处理实时视频数据;虚拟层利用虚实映射方法处理有效数据,完成物理空间与虚拟空间的虚实交互,获取实时孪生数据,建立实时选煤厂厂区三维数字孪生模型;服务层利用金字塔式场景分析网络语义分割模型,处理三维数字孪生模型,获取安全区域与生产人员轮廓边缘,按照2种轮廓边缘交并关系,完成生产人员安全视频实时定位。结果表明,该方法可有效构建选煤厂厂区三维数字孪生模型,有效分割安全区域与生产人员轮廓,提取轮廓边缘;有效检测生产人员危险行为,完成生产人员安全视频实时定位。

In order to improve the real-time positioning effect of safety video for production personnel,aiming at the problems of low definition of contour edge and difficult segmentation between safety area and production personnel contour,a real-time positioning method of safety video for production personnel in coal preparation plant based on three-dimensional digital twins was studied.The data layer collects the real-time video data of the coal preparation plant area by calling various sensors in the physical layer,and processes the real-time video data for cleaning and value-added processing.The virtual layer uses the virtual real mapping method to process the effective data,completes the virtual real interaction between the physical space and the virtual space,obtains the real-time twin data,and establishes the three-dimensional digital twin model of the real-time coal preparation plant.The service layer uses the pyramid scene to analyze the network semantic segmentation model,processes the three-dimensional digital twin model,obtains the contour edge of the safety area and the production personnel,and completes the real-time positioning of the production personnel safety video according to the intersection relationship between the two contour edges.The experimental results show that this method can effectively construct the three-dimensional digital twin model of the coal preparation plant,effectively segment the safety area and the contour of the production personnel,and extract the contour edge.This method can effectively detect the dangerous behaviors of production personnel and complete the real-time positioning of safety video of production personnel.