Learning Intelligent Factory Traffic Characteristics and Anomali Detection with Contextual Multi-Arm Slot Machine

DEĞİRMENCİ E., ÖRNEK Ö., YAZICI A.

28th Signal Processing and Communications Applications Conference (SIU), ELECTR NETWORK, 5 - 07 Ekim 2020 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/siu49456.2020.9302356
  • Basıldığı Ülke: ELECTR NETWORK
  • Anahtar Kelimeler: intelligent factories, reinforcement learning, multi arm bandit, situational awareness, anomaly detection
  • Eskişehir Osmangazi Üniversitesi Adresli: Evet

Özet

The monitoring and control of intelligent factories via data is foreseen with the Industry 4.0. Detection of abnormal events in factories by intelligent algorithms over data without human intervention is important. In this study, traffic characteristics in internal logistics are learned and anomaly detection is made through this. In the proposed method, the traffic characteristics are learned with contextual multi-arm slot machine for the intersections. Then, anomalies are detected by detecting situations that do not comply with this characteristic. The proposed method has been successfully tested for traffic data that created for simulated transport tasks in the SUMO simulation environment.