SCADA and Artificial neural networks for maintenance management

Abstract

Nowadays, the reliability of the wind turbines is essential to ensure the efficiency and the benefits of the wind energy. The SCADA system installed in a wind turbine generates lot of data that need to be processed. The information obtained from these data can be used for improving the operation and management, obtaining more reliable systems. The SCADA systems operate through different control rules that are predefined. However, a static control of the wind turbine can generate a miscorrelation between the control and the real conditions of the wind turbine. For example, two wind turbines can be separated several kilometers in the same wind farm, therefore, the operation conditions must be different and the control strategy should not be unique. This research work presents a method based on neural networks for a dynamic generation of the control strategy. The method suggests that the thresholds used for generating alarms can vary and, therefore, the control of the wind turbine will be adapted to each specific wind turbine.