基于PHM的电力中间件故障预警技术研究

Research on Fault Early Warning Technology of Power Middleware Based on PHM

摘要:

在电力信息系统中,中间件的运维工作需要从传统的事后排查提升为故障预判和智能预警,面向电力中间件的故障预测与健康管理(PHM)技术成为当前迫切需要研究的课题；分析和利用PHM技术的数据处理流程,突破性将该技术应用于智能化运维管理平台的中间件集群管理；以灰色状态下的马尔科夫预测模型为核心预测算法,结合时间切片管理和动态置信阈值技术,设计并实现了面向中间件集群管理的故障预警模型；以湖北省电力公司构建的智能化运维管理平台上的实施应用为实例,该方法实现对中间件故障准确预警,并大大降低了虚警率；实验结果证明了该方法的可行性和有效性。

Abstract:

In power information system, the maintenance of middleware should be arranged from the traditional post inspection to the fault pre-diagnosis and intelligent early-warning. Fault prediction and health management (PHM) technology based on power oriented middleware has become an urgent research topic. Taking the middleware cluster as the research object, according to the data processing flow of PHM technology, researching on Grey-Markov forecasting, combining with time slice management and dynamic confidence threshold, this paper designs and implements a fault early warning model for the middleware cluster management. Taking the application of operations management platform in Hubei electric power company as an example, The method realizes the accurate warning of the middleware fault, and reduces the false alarm rate. The experimental results show that this method is feasible and effective.