异常识别是多元统计过程控制(MSPC,MultivariateStatisticalProcessCon—tr01)方法有效应用的关键.针对现有研究对历史异常信息利用的不足,综合考虑了主成分变量贡献率与重构误差变量贡献率对异常识别的影响,将两种变量贡献率进行归一化处理并求和得到综合变量贡献率;提出了一种基于综合变量贡献率的MSPC异常识别方法,并基于matlab计算平台实现了该算法.通过田纳西过程故障模式仿真及异常识别,对该方法的应用及算法有效性进行了实例验证.%Fault detection and diagnosis is one of the key technologies on the effective application of mult- ivariate statistical process control(MSPC). In order to overcome the historical fault information using shortage, considering the influence of principal components variable contributions and the reconstructive errors, the syn- thetical variable contributions were calculated by normalizing and summing these two different variable contri- butions. A novel MSPC fault detection and diagnosis method was proposed based on the integrated variable contributions, and the relevant algorithm and program were presented and implemented. A case study was il- lustrated through the Tennessee Eastman challenge process simulation platform. The experimental results dem- onstrate that the proposed method is feasible and valid.
展开▼