火电厂锅炉常见故障的数据挖掘诊断方法

Fault Diagnosis for Boilers in Thermal Power Plant by Data Mining

针对火电厂主要大型设备之一锅炉的常见故障,提出了一种新的诊断方法:数据挖掘方法。该方法通过建立一个智能化的数据挖掘工具,直接从火电厂SCADA系统历史数据库的大量实时数据中获取故障诊断知识进行故障诊断。数据挖掘工具的核心是:采用粗糙集的约简方式,将数据库中抽取的故障诊断规则简化为基于最小变量集的决策表。由于决策表直接采用数据库中的变量来表达,有利于现场操作人员的理解与应用。该方法避免了为诊断故障而附加的专门测试或试验,降低了费用,同时减少了试验对设备造成的潜在危险。将这一方法应用于火电厂锅炉的一个复杂故障事例,结果表明:其诊断的精度可以满足现场应用的要求。

An new approach is proposed to diagnose frequent faults for boilers in thermal power plants. Based on the acquired data in SCADA（supervisory control and data acquisition）systems, a hybrid-intelligence data-mining framework is developed to extract hidden diagnosis information. The hard core of the framework is an algorithm in finding minimum size reducts which is based on rough set theory. This makes it possible to eliminate additional test or experiments for fault diagnosis which are usually expensive and involve some risks to the equipment. The decision rules mining from SCADA systems＇ database are expressed directly by points in SCADA systems＇ database, so it is easy for engineers to understand and apply to industry applications. This new approach is tested by all the data in a SCADA system＇s database of a thermal power plant, the decision rules＇ accuracy is varied from 92 % to 95 % in different months. The new approach can also be used to diagnose frequent faults for other large scale equipments in thermal power plants.