基于局部熵双子空间的多模态过程故障检测

为了提高非高斯工业过程的检测性能, 提出局部熵双子空间(LEDS)的多模态过程故障检测方法. 运用局部 概率密度估计构建数据的局部熵矩阵, 消除数据的多模态特性. 用Kolmogorov-Smirnov (KS)检验局部熵数据中变 量的正态分布特性, 对高斯分布和非高斯分布的数据分别建立基于PCA的高斯子空间和ICA的非高斯子空间故障 检测模型. 利用Bayesian决策将检测结果转化成发生故障概率的形式, 将检测结果组合成最终的统计信息, 进行故 障检测. 将该方法应用于数值例子和田纳西–伊斯曼多模态过程, 仿真结果表明, 该方法在误报率较低的情况下, 故 障检测率最高, 优于PCA、局部熵PCA(LEPCA)和局部熵ICA(LEICA)方法.

Fault detection of multimodal processes based on local entropy double subspace

In order to improve the performance of detection in non-Gaussian industrial process, a fault detection method of multimodal processes based on local entropy double subspace (LEDS) is proposed. The local entropy matrix is constructed by local probability density estimation to eliminate the multimodal characteristics of the data. The normal distribution of variables in local entropy data is tested by Kolmogorov-Smirnov (KS). The fault detection models of PCA-based Gaussian subspace and ICA-based non-Gaussian subspace are established for Gaussian distribution and non-Gaussian distribution data, respectively. The Bayesian decision is used to transform the detection results into the form of fault probability, and the detection results are combined into final statistical information for fault detection. The proposed method is applied to a numerical example and Tennessee-Eastman multimodal process. The simulation results show that the fault detection rate is the highest when the false alarm rate is lower, which is better than PCA, local entropy PCA (LEPCA) and local entropy ICA (LEICA) method.