Operational Fault Feature Extraction of Blade Based on Vibration of Wind Turbine

Abstract

By analyzing the acquisition data of the wind turbine, we calculate the correlation dimension of the vibration signals, and take it as the feature parameter which typifies the working state of blades. The calculation results indicated that the correlation dimension is effective to reflect the dynamic structure of chaotic attractor. Thus the correlation dimension of blades' vibration signals can be used to classify the different working state of blades effectively. Experiments have also shown that this method is especially effective at working state monitoring and at fault diagnosis, and can achieve higher precision in these applications.