Incipient faults are difficult to detect in plant-wide processes compared to conventional faults due to the lack of distinctive features.An incipient fault detection method based on multiple subspace weighted moving window principal component analysis(PCA)was proposed by shifting the detection perspective from the global to the local to improve the detection rate and sensitivity of incipient faults in plant-wide processes.Process variables were partitioned into different subspaces using a two-layer subspace partitioning method that combines process knowledge and data-driven approaches.A weighted moving window was used to increase the offset of incipient faults,while a local outlier factor(LOF)algorithm was introduced into PCA to further focus on the local features of the data to model fault detection in each subspace.The monitoring results in each subspace were fused with information by the Bayesian inference fusion method to obtain distributed monitoring results.The proposed method was validated by industrial examples,and the results showed that the method effectively improved the accuracy and detection speed of incipient fault detection in plant-wide processes.

With the widespread use of distributed systems, multi-subspace whole-flow industrial monitoring methods are evolving. However, due to the lack of distinctive features, incipient faults in plant-wide processes are more difficult to detect. To improve the detection rate of incipient faults in plant-wide processes while maintaining the generality of the algorithm, a novel double-layer subspace weighted moving window reconstruction independent component analysis (DS-WRICA) method is proposed. In DS-WRICA, process variables are first divided into different subspaces based on process knowledge and data-driven partitioning methods. Secondly, a weighted moving window is used to increase the offset of incipient faults, and monitoring statistics are constructed by combining optimized reconstructed independent component analysis (RICA) and local outlier factor (LOF) in each subspace. Then, the monitoring statistics in each subspace are fused with information by Bayesian inference fusion method to obtain distributed monitoring results. Finally, the effectiveness and superiority of the DS-WRICA method are verified by industrial examples.

To monitor the plant-wide process finely, a novel distributed static magnitude-dynamic difference (DSM-DD) method is proposed in this article. First, given the high dimension of the collected data in the plant-wide process, the entire data space is divided into four orthogonal subspaces according to whether the data obey Gaussian distribution and whether it has serial correlation. Second, both the static magnitude and dynamic difference of the data in the four subspaces are used to build the monitoring model. In addition, not only the features within four subspaces are extracted but the correlation between different subspaces is also extracted to construct corresponding statistics. Third, all the statistics with physical significance are put together to form a statistic vector, and the local outlier factor method is used for constructing the synthetic index to determine whether the fault occurs. Finally, the superiority of the DSM-DD method is verified through a typical industrial case.

In the industrial production, for the close-loop control, not all faults will affect product quality. To detect quality related fault effectively, a novel method named key variable-slow feature analysis (KV-SFA) is proposed in this work to extend the SFA algorithm to the domain of online quality-related fault detection. Firstly, key quality related process variables are selected via the combination of the least absolute shrinkage and selection operator (LASSO) method and the mechanism knowledge. Secondly, the SFA is conducted in the key variables space to extract slow features for establishing fault detection model. Then, the monitoring statistics are constructed and the control limits are estimated. Finally, the validity and effectiveness of the proposed KV-SFA method are proved through an industrial process.