论文部分内容阅读
为保证炼化装置生产的连续性和安全性,及时诊断和排除生产中出现的故障,提出一种基于动态贝叶斯网络(DBN)模型的故障诊断技术。采用危险性与可操作性分析(HAZOP)方法分析对炼化装置故障传播机理,详细分析系统内不同偏差发生的可能原因及可能后果,在此基础上建立复杂生产过程的故障传播动态贝叶斯模型。依据现场在线监测数据,对生产过程中出现的报警进行诊断,推理发生报警的最可能根原因。应用该方法诊断某催化裂化装置再生器温度低报警故障。结果表明,该方法诊断出的故障原因为其主风机系统故障,与事实相符。
In order to ensure the continuity and safety of the refinery and the timely diagnosis and troubleshooting of production, a fault diagnosis technique based on Dynamic Bayesian Network (DBN) model was proposed. The HAZOP method is used to analyze the failure propagation mechanism of the refinery unit, and the possible causes and possible consequences of different deviations in the system are analyzed in detail. Based on this, the fault propagation of complicated production process is established. Dynamic Bayes model. According to the on-line monitoring data, the alarm occurred in the production process diagnosis, reasoning the most likely root causes of the alarm. Application of this method to diagnose a catalytic cracking unit regenerator temperature low alarm failure. The results show that the method of fault diagnosis for the failure of the main fan system, in line with the facts.