Anomaly detection based on one-class intelligent techniques over a control level plant. (13th January 2020)

• Record Type: Journal Article
• Authors: Jove, Esteban; Casteleiro-Roca, José-Luis; Quintián, Héctor; Simić, Dragan; Méndez-Pérez, Juan-Albino; Luis Calvo-Rolle, José
• Is Part Of: Logic journal of the IGPL. Volume 28:Number 4(2020)
• Page Start: 502
• Page End: 518
• Publication Date: 2020-01-13
• Subjects: Fault detection -- one-class -- ACH -- autoencoder -- SVM; Logic, Symbolic and mathematical -- Periodicals; 511.3
• DOI: 10.1093/jigpal/jzz057
• Languages: English
• ISSNs: 1367-0751

Abstract

Abstract: A large part of technological advances, especially in the field of industry, have been focused on the optimization of productive processes. However, the detection of anomalies has turned out to be a great challenge in fields like industry, medicine or stock markets. The present work addresses anomaly detection on a control level plant. We propose the application of different intelligent techniques, which allow to obtain one-class classifiers using real data taken from the correct plant operation. The performance of each classifier is assessed and validated with real created faults, achieving successful overall results.