A hybrid prototype selection-based deep learning approach for anomaly detection in industrial machines

• Published in: Expert systems with applications 2022-10, Vol.204, p.117528, Article 117528
• Main Authors: de Paula Monteiro, Rodrigo, Lozada, Mariela Cerrada, Mendieta, Diego Roman Cabrera, Loja, René Vinicio Sánchez, Filho, Carmelo José Albanez Bastos
• Format: Article
• Language: English
• Subjects: Anomaly detection; Deep learning; Prototype selection; Rotating machinery
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2022.117528

Anomaly detection in time series is an important task to many applications, e.g, the maintenance policies of rotating machines within industries strongly rely on time series monitoring. Rotating machines are vital elements within industries. Therefore, maintenance policies on these critical elements concern the quality of products and safety issues. Condition-based maintenance is an example of those policies. In this context, we propose a novel method to train a deep learning-based feature extractor for the anomaly detection problem on rotating machinery. It consists of using a prototype selection algorithm to improve the training process of a randomly initialized feature extractor. We perform this process iteratively using data belonging to one probability distribution, i.e., the normal class. We carried the prototype selection out with the Nearest Neighbors algorithm, and the feature extractor was a Convolutional Neural Network. We validate the method on three datasets of spectrograms related to gearbox and compressors faults and achieved promising results. We obtained detection rates in anomalous data close to 100%, and the anomaly detectors classified normal instances with accuracy values superior to 95%. Those results were competitive concerning other deep learning-based anomaly detectors in the literature, with the advantage of being an integrated solution. •Learning features for anomaly detection problems may be a challenging task.•Prototype selection improves the training of feature extractors for anomaly detection.•It helps mapping normal instances to a more restricted region of the feature space.•It makes the anomaly detection via one-class classification easier.