Fault Diagnosis Method for High Voltage Trip-off of Wind Farms Based on mRMR Method and SVM

The application of large-scale integration of renewable energy technology has deepened the degree of power electronization of power grid. Therefore, the phenomenon and mechanism of wind farm trip-off is an important research issue. The major cause of wind farm high voltage trip-off are typically commutation failure fault, DC block fault and three-phase short-circuit fault and disconnection fault. In this paper, the fault diagnosis of high voltage trip-off of wind farms is investigated. Firstly, an index system is introduced according to the operation data from high voltage trip-off of wind farms. Then, the feature indicators are filtered from the index system based on Mutual Information Method and mRMR Method to identify the fault which results in high voltage trip-off of wind farms. Finally, Support Vector Machine (SVM) which is a good method for classification is applied in machine learning to verify the accuracy of fault diagnosis by these feature indicators. Actual northwest power grid is examined in case study to construct the fault scenarios for high voltage trip-off of wind farms and verify the proposed methodology.