Power electronic components are among the most crucial elements in renewable energy systems. Failure and malfunction among them can impact the reliability of such systems. Detecting anomalies in the behavior of components in the early stages can help to prevent system breakdown and financial catastrophe.

Estimating the Remaining Useful Life (RUL) of components can help to increase the reliability of systems and also reduce maintenance operations and costs, and also make it possible for logistics planning. However, as the systems become more complex, it becomes more complicated to detect or prognosis faults. Therefore, there is a need to develop more accurate and reliable diagnostic approaches which can be implemented in such systems to increase the trustworthiness of condition monitoring.

Machine Learning and Bayesian Reasoning methods are promising approaches in this regards. Furthermore, a well understanding of a system and proper modeling of it can be a great help to feature selection and extraction, and also post-processing of the results for decision-making stages.

Our purpose in this research is generalizing the current state of the art methods for wider application or developing new approaches where the current methods have shortcomings. This goal can be achieved through the following steps:

Data feature selection and extraction
Analyzing the extracted data features, and proposing an approach to perceive the relation between these features and different system status
Evaluating the proposed approach

Keywords: Condition monitoring, Predictive maintenance, Fault Detection, Fault Prognosis, Remaining Useful Life Estimation, Machine Learning, Bayesian Reasoning, Markov models, Neural Networks, Support Vector Machines, Modeling.

Funding: Self-financing.
Effektiv start/slut dato01/06/201930/06/2023


Udforsk forskningsemnerne, som dette projekt berører. Disse etiketter er oprettet på grundlag af de underliggende bevillinger/legater. Sammen danner de et unikt fingerprint.