Szczegóły publikacji
Opis bibliograficzny
Adaptive design of a cyber-physical system for industrial risk management decision support / Andrzej M. J. SKULIMOWSKI, Paweł ŁYDEK // W: ICARCV 2022 [Dokument elektroniczny] : the 17th International Conference on Control, Automation, Robotics and Vision : December 11-13, 2022, Singapore / Nanyang Technological University, Singapore. — Wersja do Windows. — Dane tekstowe. — [Piscataway] : IEEE, cop. 2022. — Dod. ISBN: 978-1-6654-7685-0 (USB); ISBN: 978-1-6654-7688-1 (print on demand). — e-ISBN: 978-1-6654-7687-4. — S. 90–97. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 97, Abstr. — A. M. J. Skulimowski - dod. afiliacja: Progress & Business Foundation, International Centre for Decision Sciences and Forecasting, Kraków, Poland
Autorzy (2)
Słowa kluczowe
Dane bibliometryczne
ID BaDAP | 144573 |
---|---|
Data dodania do BaDAP | 2023-01-11 |
Tekst źródłowy | URL |
DOI | 10.1109/ICARCV57592.2022.10004251 |
Rok publikacji | 2022 |
Typ publikacji | materiały konferencyjne (aut.) |
Otwarty dostęp | |
Wydawca | Institute of Electrical and Electronics Engineers (IEEE) |
Konferencja | 2022 17th International Conference on Control, Automation, Robotics and Vision |
Abstract
This article presents an integrated design of intelligent decision support systems (DSSs) for industrial risk management. The need for this class of systems has been necessitated by resilience building and threat response planning problems faced by complex industrial installations in energy and mining sectors. The proposed DSS software architecture is AI-based and applies causal and anticipatory networks, multi-criteria analysis, information fusion and knowledge engineering techniques. The use of AI-tools follows the AI-alignment paradigm, where AI evolution provides clues regarding the most suitable techniques to solve anticipated industrial safety problems in different time scales. We propose a general scheme of industrial risk management which includes threats, sensors, information flows, and decision-making models. This scheme is complemented by the risk optimization module, which selects the actions and actuators to implement them. Signals received from sensors are fused and confronted with threat management scenarios contained in the knowledge base. The recommendations concerning prevention, protection, and threat mitigation measures are generated by the DSS and conveyed to human decision makers for approval. Selected actions can also be autonomously initiated. Previous activity assessments result on an improved configuration of sensors and actuators, as well as more effective first responder actions. Threat and risk management modules of the DSS are linked by a sequential machine learning procedure, so that the results of prior decisions can be used to learn managerial preferences and parameters of risk mitigating scenarios.