论文部分内容阅读
针对核独立成分分析故障检测时忽略各独立成分分量对系统故障贡献度的差异,提出一种基于加权核独立成分分析的故障检测方法.使用核独立成分分析提取过程变量的独立成分,根据核密度估计衡量各独立成分分量对系统故障的贡献度,对各独立成分分量赋予不同权重,突出包含有用信息的独立成分分量,引入局部离群因子在特征空间构造统计量进行故障检测.基于数值仿真和Tennessee Eastman数据集的仿真结果表明了所提出方法的优越性.
Aiming at the difference of contribution of each independent component to system fault in the independent component analysis of fault detection, a new fault detection method based on weighted kernel independent component analysis is proposed. The independent component of the process variable is extracted by kernel independent component analysis, We estimate the contribution of each independent component to the system fault, assign different weights to the independent components, highlight the independent components that contain the useful information, and introduce the local outliers to construct the statistics in the feature space for fault detection. Based on the numerical simulation and The simulation results of Tennessee Eastman dataset show the superiority of the proposed method.