流程型生产安全数据流的多Agent节点协同分流优化方法  

Collaborative diversion and optimization method of multi-agent nodes in process-oriented production safety data flow

作　　者：张伟[1] 李泽亚 张充 赵挺生[1] ZHANG Wei;LI Zeya;ZHANG Chong;ZHAO Tingsheng(School of Civil and Hydraulic Engineering,Huazhong University of Science and Technology,Wuhan Hubei 430074,China)

机构地区：[1]华中科技大学土木与水利工程学院,湖北武汉430074

出　　处：《中国安全生产科学技术》2024年第8期5-12,共8页Journal of Safety Science and Technology

基　　金：国家重点研发计划项目(2021YFB3301100)。

摘　　要：为实现流程型生产安全监测系统的及时、准确决策,结合数据流采集与隐患识别过程,分析非关键数据的冗余、关键数据的缺失和数据计算时延较大的问题,提出基于多Agent的流程型生产安全数据流网络分流调度规划方法和节点流量划分识别机制。研究结果表明:相较于分簇传输方法,数据流网络分流调度方法可以实现关键隐患数据更高的传输成功率;相较于常规的复杂事件处理方法,本文提出的流量划分识别机制在2种类型数据集上实现隐患事件识别均有更低的计算时延。研究结果可为流程行业安全生产数字化管控模式和数据高质量获取提供参考。To achieve the timely and accurate decision-making of process-oriented production safety monitoring system,combining the data flow collection and hazard identification processes,the problems such as redundancy in non-critical data,absence of critical data,and significant delays in data computation were analyzed,and a network diversion scheduling planning method of process-oriented production safety data flow and a node flow rate partition recognition mechanism based on multiple agents were proposed.The results show that compared to cluster-based transmission methods,the data flow network diversion scheduling method can achieve a higher transmission success rate for critical hazard data.Furthermore,in comparison to conventional complex event processing methods,the proposed flow rate partition recognition mechanism exhibits lower computational latency for hazard event identification across two types of datasets.The research results can provide reference for the digitized management and control mode of work safety in process industries and the acquisition of high-quality data.

关 键 词：流程型生产 安全生产 数据流 多AGENT 数据分流 

分 类 号：X913.4[环境科学与工程—安全科学]