基于数据驱动的催化裂化过程反应-再生单元早期故障检测（英文）  

A data-driven approach for early fault detection of the reactor regenerator in the RFCC unit

作　　者：KIALAIN Kaustella 杨洋[2] 李颖[1] 张新宇[1] 杜增智[3] IFTIKHAR Maria 杨彦宇[1] 孙巍[1] 

机构地区：[1]北京化工大学化学工程学院膜分离过程与技术北京市重点实验室,北京100029 [2]顺华能源学院,辽宁石油化工大学,辽宁抚顺113001 [3]北京化工大学过程模拟优化中心,北京100029

出　　处：《计算机与应用化学》2015年第11期1304-1308,共5页Computers and Applied Chemistry

基　　金：Fundamental Research Funds for the Central Universities(ZY1511)

摘　　要：反应-再生装置是渣油催化裂化的核心部分。反应器的性能改变会直接影响产品的质量。在一个工业过程中,一般的故障都是自然或者突然产生的。大量的研究论文都致力于监测的突然故障,但发现早期故障也很重要,因为它们特点是缓慢的动态变化,最终会导致更严重的问题。建议使用的方法称为CUSUM-PCA,与传统的PCA方法相比较具有早期故障检测功能以及高级过程监控的特点。结果表明,使用CUSUM-PCA方法比传统的PCA方法提前一个小时检测出异常。The reactor-regenerator is the heart of the RFCC unit. Change in the performance of the reactor directly influences product quality. Usually faults in an industrial process can either be abrupt or incipient in nature. Significant number of research papers have been dedicated to the monitoring of abrupt faults but uncovering incipient faults is also important, as they are characterized by their slow dynamic change that eventually leads to a more serious problem. The use of a proposed method known as CUSUM-PCA is compared with conventional PCA for its early fault detection features and advanced process monitoring characteristics. Results show that the proposed CUSUM-PCA provides abnormal situation detection over an hour in advanced than conventional PCA.

关 键 词：故障诊断 积累 主成分分析 催化裂化装置 

分 类 号：TQ520.5[化学工程—煤化学工程]