中文摘要：

石油化工领域生产装置异常检测与诊断依赖机理模型和人工经验,存在主观性强、与实际参数不匹配等缺陷。在实践和总结现行石油化工生产装置异常检测与诊断技术的基础上,结合目前主流的大数据分析技术,对生产装置上获得的温度、流量、压力等数据进行实时分析,寻找大数据后面隐藏的规律,让大数据发声,及时预警和处理故障,并做出更明智的决策,减小损失。通过对某催化装置主分馏塔的塔顶温度变化幅度预警测试,比人工提前6min发现故障,将事故消除在萌芽状态。该技术具有工程实用价值,值得在工业控制现场进行实践和推广。

英文摘要：

The mechanism and model of anomaly detection and diagnosis for the production equipment in petrochemical industry have the defects of strong subjectivity and mismatch with the actual parameters.Based on the practice and summary of anomaly detection and diagnosis technology of the oil chemical production equipment,combined with the current mainstream data analysis technology,real-time analysis of production equipment on the temperature,flow and pressure data,find the hidden data behind the law,make big data sound,timely warning and fault handling,and to make more informed decisions,reduce losses.Through the test of the tower top temperature variation range of a catalytic converter,it is found that the accident is eliminated in the bud state 6 minutes ahead of time.This technology has practical value in engineering,and it is worthy to be popularized in the field of industrial control.