A system of monitoring and evaluation for performance and maintenance management in the wind turbines

Wind turbines are a great source of energy that holds much potential to deliver clean, reliable, and low-cost power to many homes and businesses. While they can reduce carbon emissions and produce clean power on a regular basis, they have a few major drawbacks, some of which may require a certain level of maintenance to the blades, bearings, and gears. This dissertation proposed a new performance evaluation system and algorithms for wind turbines to detect and predict faults that may happen in the system. This work is contrary to the currently available wind turbine monitoring and assessment systems focused on some aspects of the turbine operation, such as power output or even cost management for performance measurement. This study, however, was able to move a step further by considering other main features in the processes for evaluation, including weather conditions, wind speed, turbine loads, system vibration levels, and other factors, which gave a complete picture of the health status and performance of a wind turbine. The results implemented by the model are presented through a four-dimensional array showing how the various factors were influencing the evaluation of the performance of a wind turbine. The models were validated against two prediction accuracy metrics: The Mean Absolute Error and the Root Mean Square Error. The implementations of MAE and RMSE result in 0%, thus proving the model is accurate. The support vector regression modeling approach has already been successfully applied to fault identification in a wind turbine system. The overall accuracy in all simulations performed using the model was 99.8%, and it exhibited behavioral patterns that predicted faults and their propagation within the wind turbine system.