【作者】 赵立杰； 柴天佑；

【Author】 Zhao Lijie~(1.2) Chai Tianyou~1 ~1(Research Center of Automation , Northeastern University ,Shenyang 110004,China)~2(Shenyang Institute of Chemical Technology,Shenyang 110142,China)

【机构】 东北大学自动化研究中心；

【摘要】 过程的安全性和产品的一致性是工业界共同关注的研究课题。如果对整个间歇过程只用一个线性化模型来处理,不仅会使主元个数大大增加,而且有许多故障无法检测到。结合间歇产生过程对象特点,将整个操作周期划分成多个子过程,分别建立多个子过程MPCA模型。在监视过程中,采用事件驱动方式在多模型之间切换,然后比较子模型的SPE等统计量和统计控制线的大小,判断过程状态是否处于异常。根据贡献图识别故障发生的原因,定位故障源。将所提出的建模和监视策略应用到锦西化工集团聚氯乙烯间歇生产,该方法能够快速有效地监视PVC生产过程中每一阶段性能的变化,及时准确地识别故障和诊断故障发生的原因,同时模型内存空间占用少,计算量小。更多还原

【Abstract】 Process safety and good quality are the common objectives in industries. If a single linear MPCA model is used to characterize the entire batch process ,PC number may increase and some faults may not be identified. According to the features of multistage operations,batch process can be divided into a series of sub-process, which is built a local MPCA model for each stage. Several key events switch the MPCA model from one stage to another when the batch is monitored. The proposed method is applied to the PVC process in Jinxi chemical plant, which can quickly found abnormal change of each stage and correctly identified the cause of fault. It also reduces memory space and shorts computation time.更多还原