Identification of Offshore Wind Farms Electrical Abnormal State Based on Multi-dimensional-matrix Profile

As the increasing installed scale of offshore wind farms, the harsh environment and the complexity of equipment lead to frequent occurrence of electric abnormal states in offshore wind farms. However, the lack of sufficient abnormal state samples in offshore wind farms makes it difficult for traditional identification methods to achieve accurate online identification of abnormal state. Therefore, this paper proposes a method for identifying the electric abnormal states of offshore wind farms based on multi-dimensional-matrix profile (MDMP) algorithm, which can realize remote monitoring and online diagnosis of the operating status of offshore wind farms. First, the ensemble empirical mode decomposition (EEMD) algorithm is used to effectively mine the fault and disturbance historical data of the offshore wind farms, and to extract the features to construct the feature sample library of abnormal states without training process. Then, real-time data of abnormal operation of offshore wind farms are obtained, and feature extraction is performed. Finally, the MDMP method is used to match the real-time abnormal sample features with the abnormal sample library to realize the abnormal state identification. In addition, considering the computational burden in reality, a heartbeat packet mechanism is introduced to detect electrical abnormal waveforms in offshore wind farms, which can effectively save computing resources. The effectiveness and scalability of the identification method are verified by Matlab/Simulink simulation and actual engineering data.