炼化过程故障传播路径的不确定并行推理方法研究

Uncertain parallel reasoning method for fault propagation in refinery process

为有效抑制炼化过程故障传播,基于模糊Petri网(FPN)的基本原理,采用FPN网建立炼化过程故障传播模型。对每次发生变迁的各故障库所进行不确定并行推理,将每次变迁发生后所得故障发生概率的最大值作为其后集库所对应命题成立的可信度,故障依次向可信度值最大的后集库所传播,通过故障的前后集关系得出最可能的故障传播路径。将此方法应用于典型炼化过程的常压塔装置。结果表明,引起常压塔故障的最可能传播路径为:进料温度异常—常压塔塔底温度异常—产品质量不合格—常压塔故障。在该路径中的常压塔底部设置温度报警器,进行重点监控,便能保证炼化过程安全可靠进行。

In order to determine and prevent fault propagation in refinery process, based on FPN mecha- nism, a uncertain parallel reasoning method was worked out for fault propagation in refinery process. A model was built by using FPN for fault propagation in refinery process. On the basis of uncertain data given by experts, uncertain parallel reasoning was made for faults of every change. The back event was given by the largest fault probability after every change, and the fault propagated to the back event. The most proba- ble fault propagation path was obtained through the relations of faults. Atmospheric tower device in typical refinery process was taken as a research object. The uncertain parallel reasoning method was used for stud- ying atmospheric tower fault propagation. The results show that the most probable fault propagation path for atmospheric tower is： abnormal feed temperature, atmospheric tower bottom＇s abnormal temperature, un- qualified product quality, atmospheric towerg failure, and that both installing a temperature alarm at atmos- pheric tower bottom of the path and carrying on key monitoring, will ensure safe and reliable operation of tower.