基于动态限的周期非稳定工况的实时故障检测模型

Real-Time Fault Detection Model Based on Dynamic Limit under Periodic Non-Steady Condition

系统非稳定工况下,传统的多元统计方法难以对多变量参数进行有效的故障检测。针对这些问题,本文给出了动态限和同峰谷的概念;并证明了周期非稳定工况下的多变量参数的T2统计量也是周期性的;然后提出了一种基于动态限的周期非稳定工况的实时故障检测模型,并对模型的实时性和可行性进行了分析。最后将该模型应用于系统周期非稳定工况下的实时故障检测。

The traditional multivariate statistical methods could not detect the fault effectively under the non-steady condition of complex system. For these problems above, the concept of the dynamic limit and the same peak-valley are given in this paper. And, the paper proves that the T2 statistics of the multivariable parameters is periodic if the system is under the periodic non-steady condition. Then, this paper proposes a model of real-time fault detection under the periodic non-steady condition and gives the real time and the feasibility analysis of the model. At last, the model is applied to the periodic non-steady conditions in real time.