Thesis supervisor: Zoltán Süle
co-supervisor: Tamás Ruppert
Location of studies (in Hungarian): University of Pannonia, Department of Computer Science and Systems Technology (MIK), Department of Systems Engineering (MK) Abbreviation of location of studies: PE
Description of the research topic:
The research aims to develop risk intelligence-based predictive models that provide early intervention strategies for systems with uncertainty. The planned research program aims to identify and forecast the risks of systems and processes using data-driven artificial intelligence and machine learning algorithms. During the research, we will explore and introduce new methods on how data science, artificial intelligence, machine learning, and network science can be applied to uncover and characterize risk factors, identify correlations, and learn and act based on relevant information. The research will focus on the accuracy of machine learning models and the correctness of decisions supported by these models. The information extracted by risk intelligence-based predictive systems will be applied in scenario analyses, optimization, and multi-objective decision-making.
The key steps in developing risk intelligence solutions during the research work are as follows: (1) Recognizing learnable risks: uncovering hidden correlations and identifying measurable and identifiable dependencies. (2) Ranking risks based on a learning process: predicting outcomes, generating early warnings, and scheduling and fine-tuning machine learning tasks considering related risks. (3) Optimizing actions and decisions: executing simulations for various scenarios and alternative outcomes, conducting "what-if" analyses, and distributing risks within risk networks.
The research work should focus on the targeted development and application of the following theoretical background:
- machine learning algorithms (neural networks, deep learning, event analysis);
- survival analysis-based solutions, parametric models;
- optimization procedures (mixed-integer linear and non-linear optimization solutions);
- elements of artificial intelligence;
- process mining algorithms;
- process simulation tools;
- process informatics solutions (related standards, ontologies).
Preliminary results can be found in the following publications:
[1] Süle Z, Baumgartner J, Dörgő G, Abonyi J.: P-graph-based multi-objective risk analysis and redundancy allocation in safety-critical energy systems, Energy, 179, 989-1003. (2019)
[2] Bakon K., Holczinger T., Süle Z., Jaskó Sz., Abonyi J.: Scheduling under uncertainty for Industry 4.0 and 5.0, IEEE Access, 10, 74977-75017. (2022)
[3] Czvetkó T., Kummer A., Ruppert T., Abonyi J.: Data-driven business process management-based development of Industry 4.0 solutions. CIRP journal of manufacturing science and technology, 36, 117-132. (2022)
[4] Grimstad J., Ruppert T., Abonyi J., Morozov A.: Preventive Risk-based Maintenance Scheduling using Discrete-Time Markov Chain Models, Proc. of the 33rd European Safety and Reliability Conference (ESREL 2023), doi: 10.3850/978-981-18-8071-1_P367-cd (2023)
Further achievements: The related R&D projects of the University of Pannonia, the research of the Department of Computer Science and Systems Technology, and the Department of Systems Engineering, as well as the achievements of the MTA-PE Research Group on Monitoring Complex Systems, are available at http://www.abonyilab.com.
Login
