王美林 黄钧

摘要：现有的车间大型设备复杂度高，越来越依赖监控系统对其进行实时性管理。目前依靠云计算的设备监控系统所需传输的数据量越来越大，导致需求紧急响应的告警信息容易出现延迟，严重会造成设备停机等问题。针对现有PVC压延车间设备告警方案定位不明确和告警响应不及时的问題，该文引入边缘计算概念，设计了一个新型设备监控系统，将告警事件处理放置在边缘端，减轻数据长距离传输带来的高延时性和云计算中心处理压力，利用用户定义的阈值告警机制和特征树去重策略减少告警频繁带来的运维困难，提高设备告警处理针对性和实时性。通过对比分析，加入边缘计算对监控系统的告警实时性有了较大的提升。

关键词：边缘计算;PVC车间;阈值告警;去重;监控

中图分类号：TP206+.3      文献标识码：A

文章编号：1009-3044（2021）15-0218-04

Abstract： The existing workshop large-scale equipment has high complexity， more and more rely on the monitoring system for its real-time management. At present， the amount of data needed to be transmitted by the equipment monitoring system relying on cloud computing is increasing， which leads to the delay of alarm information requiring emergency response， and seriously causes equipment downtime and other problems. Aiming at the problems of unclear positioning and untimely alarm response of existing PVC calendering workshop equipment alarm scheme， this paper introduces the concept of edge computing and designs a new equipment monitoring system， which places the alarm event processing at the edge end to reduce the high delay caused by long-distance data transmission and the processing pressure of Cloud Computing Center， and uses the user-defined threshold alarm mechanism and feature tree to remove the alarm Heavy strategy reduces the operation and maintenance difficulties caused by frequent alarms， and improves the pertinence and real-time of equipment alarm processing. Through comparative analysis， the real-time alarm performance of the monitoring system is greatly improved by adding edge computing.

Key words： edge computing; PVC workshop; threshold alarm; deduplication; monitoring

1 背景

随着工业4.0建设的不断推进，我国传统工业信息化进程飞速发展，车间的大型机械设备日益呈现复杂化、大型化和多功能化[1]。同时，针对设备的各项数据采集无论从数量上或者精度上也有了质的飞跃，这对设备的监控也提出了新的问题，如何加速数据的实时性处理使得故障信息更快地响应也是车间监控告警的迫切需求。

针对现有大型设备监控系统数据的复杂性和关联性，张棋胜在云计算平台监控系统的研究与应用[2]中设计了推拉混合式数据采集算法以及在云端建立预测模型进行数据处理，大大提高了数据处理的质量。……