摘要
针对虚拟空间与实际环境存在偏差,容易造成监测结果与真实场景脱节,此外,虚拟空间中通常采用特定的建模格式和表示方法,直接将其作为AI的输入数据可能导致可监测变化指数的数值水平较高等问题,提出基于AI与数字孪生技术的高空作业安全监测方法。首先,利用数字孪生技术建立高空作业环境虚拟模型,在该模型的服务单元中添加弹塑性材料参数,可准确地模拟高空作业环境的实际状况,提高安全监测的准确性和可靠性。然后,通过同步映射技术,将数字孪生虚拟空间转化为数字空间,为AI算法提供更精准的数据输入。最后,在数字空间中,利用AI技术中的单发多框检测器(single shot multibox detector,SSD)算法定位高空作业面中的作业人员,根据头部及脚部的关节点判定当前的安全状态,实现高空作业安全监测过程。根据实验结果可知,经该方法采集并处理高空作业场景后,可以精准定位到高空环境中的作业人员,且可监测变化指数水平较低,可以满足高空作业保障需求。
Due to the deviation between the virtual space and the actual environment,it is easy to cause the monitoring results to be disconnected from the real scene.In addition,specific modeling formats and representation methods are usually used in virtual space,and directly using as the input data of AI may result in a high value level of the monitorable change index.In this regard,a safety monitoring method for aerial work based on AI and digital twin technology is proposed.Firstly,the virtual model of aerial work environment is established by using digital twin technology,and the elastic-plastic material parameters are added to the service unit of the model,which can more accurately simulate the actual situation of aerial work environment and improve the accuracy and reliability of safety monitoring.The digital twin virtual space is transformed into digital space through synchronous mapping technology,providing more accurate data input for AI algorithm.In the digital space,the single shot multibox detector(SSD)algorithm in AI technology is used to locate the operators in the aerial work surface,and the current safety state is determined according to the joint points of the head and feet,so as to realize the safety monitoring process of aerial work.According to the experiment,this method can accurately locate the operators in the high-altitude environment after collecting and processing the high-altitude operation scene,and the monitoring change index level is low,which can meet the needs of high-altitude operation support.
作者
曾懿辉
张虎
麦俊佳
ZENG Yihui;ZHANG Hu;MAI Junjia(Foshan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Foshan 528000,Guangdong,China)
出处
《自动化技术与应用》
2026年第4期87-91,共5页
Techniques of Automation and Applications
基金
中国南方电网公司科技项目(GDKJXM20230890)。
关键词
高空作业
安全监测
数字孪生技术
单发多框检测器算法
人工智能
high altitude operations
safety monitoring
digital twin technology
SSD algorithm
artificial intelligence
