Fault Detection Method Based on k-Nearest Neighbor Normalization and Weight Local Outlier Factor for Multimode Process Circulating Fluidized Bed Boiler | Chapter 10 | Research and Developments in Engineering Research Vol. 1

 Mode changes in contemporary's sophisticated industrial processes influence unanticipated shutdowns, that could reduce the age of critical system and result in high support costs. A technique that can recognize the fault of equipment operating in miscellaneous modes is wanted to prevent this issue. In light of this, we suggest a novel mistake detection method that makes use of the k-nearest neighbor normalization-located weight local eccentricity factor (WLOF). The suggested approach completes activity local normalization using neighbors to take into account potential way changes in the normal dossier, and WLOF is used for fault discovery. In contrast to statistical forms, such as principal component analysis (PCA) and free component analysis (ICA), the local aberration factor (LOF) uses the density of neighbors. However, common LOF is greatly afflicted by the distance between allure neighbors, performance can deteriorate when faulty dossier is adjacent to usual data. To improve the defect discovery performance of LOF, the proposed plan multiplies proportionally by the distance middle from two points each neighbor. A circulating fluidized bed boiler and a multimode mathematical case were used to assess the influence of the suggested method. The results of the experiments show that the suggested order performs better than conventional PCA, essence PCA (KPCA), k-nearest neighbor (kNN), and LOF. In particular, distinguished to conventional techniques, the submitted method raised detection accuracy by 20%. Therefore, the projected technique is appropriate to a real process operating in multiple fads.

Author(s) Details:

Minseok Kim,
Department of Electrical and Electronics Engineering, Pusan National University, Busan 46241, Korea.

Seunghwan Jung,
Department of Electrical and Electronics Engineering, Pusan National University, Busan 46241, Korea.

Baekcheon Kim,
Department of Electrical and Electronics Engineering, Pusan National University, Busan 46241, Korea.

Jinyong Kim,
Department of Electrical and Electronics Engineering, Pusan National University, Busan 46241, Korea.

Eunkyeong Kim,
Department of Electrical and Electronics Engineering, Pusan National University, Busan 46241, Korea.

Jonggeun Kim,
Artificial Intelligence Research Center, Korea Electrotechnology Research Institute, Changwon 51543, Korea.

Sungshin Kim,
Department of Electrical and Electronics Engineering, Pusan National University, Busan 46241, Korea.

Please see the link here: https://stm.bookpi.org/RADER-V1/article/view/10096