On the combination of support vector machines and segmentation algorithms for anomaly detection: A petroleum industry comparative study

摘要: Abstract Anomaly detection has to do with finding patterns in data that not conform an expected behavior. It recently attracted the attention of research community because its real-world application. The correct unusual events empower decision maker capacity act on system order correctly avoid, correct, or react situations associated them. Petroleum industry is one such application scenarios. In particular, heavy extraction machines for pumping and generation operations like turbomachines are intensively monitored by hundreds sensors each send measurements a high frequency damage prevention. For dealing this lack labeled data, paper we describe combination fast quality segmentation algorithm one-class support vector machine approach efficient anomaly turbomachines. As result perform empirical studies comparing our another using Kalman filters real-life related oil platform turbomachinery detection.