文献类型：期刊文献

中文题名：基于模糊有序聚类算法的间歇过程故障检测

英文题名：Fault Detection for Batch Processes Based on Fuzzy Order Clustering Algorithm

作者：张成[1];赵海涛[1];孙韶媛[2]

机构：[1]华东理工大学信息科学与工程学院,上海200030;[2]东华大学信息科学与技术学院,上海201620

年份：2018

卷号：44

期号：6

起止页码：940

中文期刊名：东华大学学报（自然科学版）

外文期刊名：Journal of Donghua University(Natural Science)

收录：CSTPCD;;北大核心:【北大核心2017】；CSCD:【CSCD_E2017_2018】；

基金：国家自然科学基金资助项目(61375007);上海市科委应用基础研究资助项目(15JC1400600)

语种：中文

中文关键词：多阶段;数据不等长;间歇过程;模糊有序聚类算法;故障检测

外文关键词：multi-stage; unequal length of data; batch process; fuzzy order clustering algorithm; fault detection

摘要：针对间歇过程故障检测中的断续以及数据不等长等问题,提出了一种模糊有序聚类算法(fuzzy order clustering algorithm,FOCA),实现了多阶段间歇过程的故障检测。FOCA在对过程数据进行子阶段划分时加入模糊策略,在初始化聚类时严格保证类内数据连续,且只针对单独的批次数据进行聚类,有效地解决断续、数据不等长等问题,优化了子阶段划分的结果。将FOCA应用于青霉素发酵过程的故障检测中,仿真结果表明该方法有效地降低了故障漏报率以及误报率。
Aiming at the problem of random order and unequal data in batch process fault detection,a fuzzy order clustering algorithm(FOCA)was proposed to implement batch processes monitoring.A fuzzy strategy was utilized when FOCA was performed on batch process phase partition.The clustering data was guaranteed to be continuous within the initial cluster.And FOCA was used to cluster only for individual batch.Thus,FOCA solved the problem of discontinue,unequal length of batch processes effectively,and the results of phase partition were optimized.This method was applied to the fault detection of penicillin fermentation process.Simulation results show that this method reduces the missing alarm and false alarm rate effectively.