Interaction between Machine Learning and Maintenance Planning for Wind Energy Systems: A Theoretical Review and Analysis

Maintenance planning for wind energy systems attempts to reduce downtimes under operating conditions that could leave possible damages ranging from loss of reputation to financial and potential losses of life. Today, with advanced sensing technologies, machine learning plays an important role and becoming dominant in the field of condition monitoring (i.e. detection, diagnosis, and prognosis) while assessing health conditions timely. By using specific methods according to specific scheduling criteria, results of condition monitoring systems are considered very useful in predicting the required maintenance tasks and their appropriate timing. More simply, maintenance schedule is constructed by studying the interactions between machine learning predictions and maintenance strategies. To the best of authors’ knowledge, there are no theoretical studies including specific review papers made so far introducing such a topic. Against this background, this paper is dedicated to exploring related review papers from a general perspective, analyzing research gaps, and providing insights required to craft a future comprehensive review paper specifically dedicated to exploring such a subject. More attention is paid to the theoretical point of view, as it tends to pave the way for future research on the subject. The related review papers are therefore analyzed based on how machine learning and maintenance planning are viewed from the perspective of condition monitoring and task scheduling. Examination of the results reveals important gaps in research, while suggesting future possibilities for practical, and effective comprehensive review study.