基于复杂事件处理的RFID辅助实时生产监控

RFID integrated real-time manufacturing monitoring based on complex event processing

将RFID(radio-frequency identification)技术应用于生产现场监控有2个关键:信息实时性,即如何将大量标签原始数据高效地转化为对企业有用的复杂事件信息;逻辑复杂性,即如何适应监控业务逻辑的复杂多变。目前缺乏系统性的设计和实现方法兼顾这2点,因此该文面向实时生产监控中RFID技术的各种应用模式,提出一种RFID实时事件处理系统的设计方法及其实现案例。一方面提供一套形式化的复杂事件建模语言以编程式的方法定义源自生产监控任务的事件逻辑,支持复杂的时间和空间语义,并提供事件层级体系降低建模复杂度,提高重构效率;另一方面基于复杂事件处理(CEP)等方法实现复杂事件的提炼和推理,及时挖掘潜在的生产状态或异常。据此开发的原型系统,在生产现场监控实践测试中表现出较好性能。

RFID（radio-frequency identification） integrated manufacturing monitoring aims to transform primitive RFID events,which are collected in real time,into useful information（complex events）for managers or higher level IS.The process must be efficient and reconfigurable to tackle real-time,complex monitoring tasks.This paper describes an architecture for designing and implementing an RFID real-time event processing system which adapts to manufacturing monitoring applications.A formal complex event modeling language with temporal and space semantics is provided to tailor the monitoring task logic by writing simple scripts.The event hierarchical model also reduces the modeling complexity.Complex event processing（CEP）can then be used for complex event filtering and reasoning.Hidden manufacturing issues and exceptions can thus be detected in a timely manner.A prototype system was developed and tested in a realistic manufacturing scenario.