项目名称: 石化复杂过程非正常工况实时故障诊断方法研究
项目编号: No.51205340
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 机械工程学科
项目作者: 郭丽杰
作者单位: 燕山大学
项目金额: 25万元
中文摘要: 石油化工装置的大型化、集约化和一体化给行业安全生产带来新的挑战,一旦发生事故,后果严重,影响范围大,如何确保其安全生产是亟待解决的重要问题。本研究从安全系统工程的理论和角度出发,对石化装置非正常工况的实时故障诊断预警方法进行研究。首先,建立基于动态模拟的定量化危险与可操作性分析方法,描述事故发生、发展和爆发全生命周期的过程,研究事故演变机理。在此基础上,对定量化危险与可操作性分析及动态模拟结果数据进行主元统计分析,提取故障征兆-原因数据特征,建立数据库。然后,采用动态轨迹法对装置实时监测数据进行比较-判别分析,辨识非正常工况;将所建立的数据库数据作为训练样本,建立神经网络故障诊断模型,以实现装置的实时故障诊断预警。最后,利用精馏实验装置验证所提出理论和方法的正确性。本研究致力于石化装置非正常工况的管理,重点研究正常工况与非正常工况之间的转化机理,并且提出非正常工况监测、故障诊断决策的方法。
中文关键词: 危险与可操作性分析;非正常工况;过程监测;故障诊断;石化装置
英文摘要: The petrochemical plant of large-scale, intensive and integration has brought new challenges to production safety of the industry. The accident will result in serious consequences and large influence range once the failure of the plant occurs. How to ensure the safe production is an important issue to be solved. Based on theory of safety system engineering, real-time fault diagnosis and early-warning method of abnormal situation on the petrochemical plant will be investigated in the study. Firstly, a dynamic simulation based Hazard and Operability analysis (HAZOP) methods will be established with aim to describe the full life cycle process including the accident occurrence, development and outbreak of accident. And the accident evolution mechanism will also be investigated. On the basis of this method, the quantitative HAZOP and dynamic simulation results data will be analyzed statistically by Principal Component Analysis method so that the symptoms-causes of fault can be extracted and the database will be established. The data from real-time monitoring will be compared with those from prior process and judged by Dynamic Locus Analysis method for identification abnormal process state. Neural network model will be established in which the database can be taken as training samples. As a result, real-time fault di
英文关键词: hazard and operability analysis;abnormal condition;process monitoring;fault diagnosis;petrochemical plant