XGEN 2024 - International Conference of Scientific Communications

Working machines are largely used in industrial environments for production of items and generate lots of data following these processes. The proper operation of machines influence the production output, thus detection of anomalies in machines activities is a crucial thing for avoiding awful outcomes. This paper present an ensemble unsupersived anomaly detection method able to handle aspects such as efficiency and data volume. Proposed method consists of 2 stages: in the first stage, statistical-based methods are used to assign labels to input data, then second stage use feature bagging technique to create and train estimators later used for prediction.