Identification of Multiple Sensor Disturbances during Process Monitoring

Autor:	Bruce R. Kowalski, Chris L. Stork, and David J. Veltkamp
Rok vydání:	1997
Předmět:	business.industry Chemistry Process (computing) Univariate Pattern recognition Fault (power engineering) Residual Analytical Chemistry Identification (information) Sensor array Statistics Principal component analysis Process control Artificial intelligence business
Zdroj:	Analytical Chemistry. 69:5031-5036
ISSN:	1520-6882 0003-2700
Popis:	A novel, automated method based on principal component analysis is presented for the detection and identification of disturbed sensors during a process monitoring application. As opposed to previous approaches, which are capable of identifying a fault in only a single sensor, the backward elimination sensor identification (BESI) algorithm is presented, which can identify upsets in multiple sensors. In the method, disturbed sensors are identified sequentially, or one at a time, using a residual-based criterion. The BESI algorithm is sensitive to changes in sensor correlations detected by conventional multivariate statistical process control but offers the ease of interpretation of conventional univariate methods. The BESI algorithm is successfully employed in the identification of disturbed sensors in simulated spectroscopic data and industrial process data. By examining the disturbance profiles generated over time by the BESI algorithm, it is also possible to distinguish between sensor and process disturb...
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::5413e99676bd719ad1e48e8f2adf62c4 https://doi.org/10.1021/ac9704366 Zobrazit plný text záznamu