earticle

영어

In recent years, many countries have actively studied wind power generation as a means of realizing low-carbon green growth through a new renewable energy source. The most efficient method of securing the stable operation of wind turbines and reduce maintenance costs is monitoring and analyzing their operational status in realtime through a remote monitoring system. Remote monitoring systems employ various sensor technologies and the Wireless Sensor Network to collect and transmit data on the status of individual parts in realtime, and they diagnose faults through a signal analysis system. Application of the fault analysis method can reduce fault resolution times and minimize losses. In this study, signals collected from wind turbines were analyzed, and their characteristics were extracted through empirical mode decomposition (EMD). In the experiment, EMD learning was carried out using the following fault signals as examples: The back-propagation (BP) neural network algorithm with generator vibration, an unbalanced rotor, and a bearing misalignment fault. This article proposes a method of diagnosing faults through signal analysis and recognition, and it demonstrates the validity of the method through a simulation.