ML-based Fault Tree Analysis of Industrial Router Reliability in the System IIoT

• Published in: 2024 International Conference on Applied Mathematics & Computer Science (ICAMCS) pp. 173 - 178
• Main Authors: Kolisnyk, Maryna, Jantsch, Axel
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.09.2024
• Subjects: Fault diagnosis; Fault trees; FTA; IIoT; Industrial Internet of Things; Probability; Production; Python; reliability; router; Software; Software algorithms; Statistical analysis; System analysis and design
• DOI: 10.1109/ICAMCS62774.2024.00028

For today's complex and expensive systems, such as an industrial router in an Industrial Internet of Things (IIoT) network, it is important to ensure high operational reliability. The router is a complex system consisting of many subsystems, and an important task is to identify possible failures and faults of its subsystems in order to prevent them in time, or to parry and eliminate them. One of the methods of system reliability assessment is the method of cutting, which includes fault tree analysis (FTA). FTA allows, using the obtained and analysed statistical data, to build a fault tree with a given level of detail, and to determine the probabilities of failures of one or another subsystem. In this paper a detailed tree of failures and failures of hardware (HW), software (SW), and hardware-software (HW-SW) of the industrial router of the IIoT system is developed. The tree is built with the use of program code, using ensemble supervised ML algorithms in Python. The obtained graphical dependencies and obtained probabilities of failures of router's subsystems as a part of quantitative analysis, allowed us to determine for the given initial data the most exposed to the risk of failure of the subsystem of the industrial router in the system IIoT.