Estudo preliminar para desenvolvimento de um sistema de manutenção em turbinas eólicas de uma planta híbrida, aplicando técnicas de IA, para clusterizar e classificar dados.

Abstract

In recent years there has been an increase in wind generation, driven by environmental factors and the incentive offered for the development of clean and sustainable technologies for energy generation. However, due to the rapid growth of this technology, concerns about the safety and reliability of wind turbines are increasing, especially due to the associated risks and financial costs. Therefore, health monitoring and fault detection for wind turbines has become an important research focus. Thus, the objective of this work was to development of a maintenance system for wind turbines of a hybrid plant, realizing initial exploratory studies, starting from a technological prospection on tools and methods for detecting faults in wind turbines and followed by the construction of a computational model to analyze historical data from CHESF wind farms, using AI tools, to look for patterns and classify this data. For this, documents was analised and a technological prospection was made on techniques and tools developed to detect faults in wind turbines, in addition to the application of artificial intelligence as an innovation; data from operation and maintenance reports of CHESF’s wind turbines were collected and analyzed, looking to check patterns and correlations; Finally, a computational model was built, using AI tools, to cluster and classify the data. Interesting results were achieved by analyzing the data with grouping and classification tools. In the clustering of the data, 3 clusters were found, which indicate possible behaviors of: normal operation, pre failure and failure, but is required deeper analyzes, using more datasets and a detailed maintenance mapping, to ensure more conclusive results. The model built for training and recognition of the clusters obtained accuracy above 98%. It is important to extend the analyzes using new input datasets to be sure of the model’s effectiveness. These studies will be extended and should contribute to the development of a maintenance system for wind turbines at the hybrid plant. The differential of this work is the use of real data from CHESF’s wind turbines, in addition to the knowledge of the maintenance teams, combined with new AI technologies and scientific support from Senai Cimatec professors. Another important contribution is in relation to permanent magnet wind turbines, as there are not many studies in this field, therefore a great potential to be explored.