董锴文, 孙彦景, 陈岩, 云霄, 程小舟, 王博文, 周玉, 王斌. 矿山生产作业场景视频结构化分析关键技术[J]. 煤炭学报, 2021, 46(11): 3724-3735.
引用本文: 董锴文, 孙彦景, 陈岩, 云霄, 程小舟, 王博文, 周玉, 王斌. 矿山生产作业场景视频结构化分析关键技术[J]. 煤炭学报, 2021, 46(11): 3724-3735.
DONG Kaiwen, SUN Yanjing, CHEN Yan, YUN Xiao, CHENG Xiaozhou, WANG Bowen, ZHOU Yu, WANG Bin. Key technology of video structured analysis in mine production operation scenario[J]. Journal of China Coal Society, 2021, 46(11): 3724-3735.
Citation: DONG Kaiwen, SUN Yanjing, CHEN Yan, YUN Xiao, CHENG Xiaozhou, WANG Bowen, ZHOU Yu, WANG Bin. Key technology of video structured analysis in mine production operation scenario[J]. Journal of China Coal Society, 2021, 46(11): 3724-3735.

矿山生产作业场景视频结构化分析关键技术

Key technology of video structured analysis in mine production operation scenario

  • 摘要: 随着5G通信、边缘计算、大数据和人工智能等技术在煤炭工业生产场景的应用,各类视频设备和功能系统急剧增多。为提高矿山安全生产的协同管控能力和智能决策水平,解决矿山各业务系统智能化程度低、深度信息感知能力差以及决策能力弱的问题,需要对具体生产场景下的视频结构化分析相关技术进行深入研究,构建跨场景的态势感知与协作、全流程的生产状态监控的智慧矿山生产视频结构化分析系统平台。建议从人、机、料、法、环5个维度融合分析矿山视频数据,通过场景认知和生产规则建模以应对复杂作业场景下业务场景识别的需求,对矿山生产业务及规则进行结构化特征建模,从时间、空间以及语义等方面对目标特征进行组合分析与分级索引,在云边协同技术架构下,让感知数据上行、规则模型下行,研发能够快速进行全系列结构化数据传递、全场景语义信息融合以及全流程生产风险管控的云边协同视频结构化分析平台,形成更加符合矿山工业形态的非结构化数据治理模式。在深入分析矿山安全生产需求和视频结构化分析技术特性的基础上,提出研究矿山作业场景视频结构化分析机制、矿山作业场景认知与规则建模技术、人员安全行为分析督导技术、设备状态视频监控联合分析技术、视频分析应急协同与联动技术等关键技术,对建立集人机督导和生产监控于一体的智慧矿山生产视频结构化分析系统,促进矿山安全生产系统由局部信息化、自动化向多系统协同智能化发展具有重要意义。

     

    Abstract: With the increasing applications of technologies such as 5G communication, edge computing, big data, and artificial intelligence in the coal mine production scenarios, the types of video equipment and functional systems have dramatically increased.To enhance the capacity of collaborative management and control and the level of intelligent decision-making for mining safe production, as well as to simultaneously solve the problems of low intelligence level, poor perception for deep information, and weak decision-making ability in all business systems throughout the mining production process, it is necessary to study the intelligent application of the video structured technology in different production scenarios in-depth, and establish a video structured analysis system for the intelligent mine with the abilities of cross-scenario situational awareness and collaboration, as well as the status monitoring of whole production process.This paper suggests to analyze the video data from five dimensions, i.e., person, machine, material, regulation, and circumstance, and then proposes to utilize the scenario recognition and rule modeling technologies to meet the business requirements of complex mine scenarios.Furthermore, the structural features of mine production business and rules are modeled to comprehensively analyze and hierarchically index the target features in terms of time, space and semantics.Based on the cloud-edge collaboration architecture, a cloud-edge collaboration video structured analysis platform for full-type structured data transmission, full-scenario semantic information fusion, and full-process production risk management is studied.This platform, empowered with the perception data upload and rule models download, is able to construct an unstructured data governance mode that conforms to the mine characteristics.Based on the in-depth analysis of mine safe production requirements and the technical characteristics of video structured analysis, this paper proposes to study the key technologies including video structured analysis mechanism, cognitive and rule modeling technology, personnel safety behavior analysis and supervision technology, equipment status joint analysis technology, and emergency coordination and linkage technology in the mine operation scenarios.These technologies are of great significance to build the mining video structural analysis system with the functions of both human supervision and production monitoring and boost the development of mining production systems from the local informatization to the collaborative intelligence of multi-system.

     

/

返回文章
返回