摘要
主元分析法(PCA)和部分最小二乘法(PLS)是目前应用较普遍的故障检测方法,但是这两种方法都不能解决变量中存在自相关的问题,而化工过程测量变量之间往往存在互相关和强自相关;规范变量分析(CVA)法能有效地解决上述问题,应用于过程故障检测的效果明显优于PCA和PLS;但目前的研究大都限于故障检测上,在此基础上进行故障诊断的研究很少,因此提出了结合CVA统计模型和SPE统计量,再利用贡献图的方法,实现对Tennessee—Eastman过程的故障诊断;仿真结果表明了所提出方法的有效性。
Principal component analysis （PCA） and partial least squares （PLS） are widely used in fault detection at present, but they all cannot deal with auto-correlation problem which exists in variable, However, most chemical processes variables have cross-correlation and strong auto-correlation, Canonical variate analysis can solve the above problem effectively, and the result of fault detection is superior to PCA and PLS clearly when CVA is applied to practical processes. But the present researches mostly limit to fault detection. The research of fault diagnosis based on detection is very few. Combining CVA statistical model with SPE statistic and using contribution bar, this paper achieves fault diagnosis of Tennessee-Eastman process. Simulation results show the effectiveness of the proposed method.
出处
《计算机测量与控制》
CSCD
2007年第8期984-986,993,共4页
Computer Measurement &Control
基金
新世纪优秀人才支持计划(NCET-05-0485)。
