Abnormal Wind Power Data Identification Based on the Improved FCM Algorithm and Considering the Influence of Wind Speed

In the big data era, the planning, operation and maintenance of power systems are increasingly dependent on the support of various power data. Wind turbine operating data is an important data set considering the increasing installation capacity worldwide. However, due to factors such as unscheduled shutdown, load shedding and communication noise, a great variety and amount of abnormal data may be present in wind power data, which harms the economic and safe operation of wind turbines. In some cases, abnormal operating status even cannot be detected in time due to the disruption of abnormal data, causing serious accidents. Therefore, it is necessary to identify the abnormal wind power data from massive measurements and ensure the availability of accurate and effective data. In this paper, an abnormal wind power data identification strategy is proposed by the improved fuzzy c-means (FCM) algorithm and considering the influence of wind speed. Specifically, the feasible domain matrix is employed to identify abnormal data while the mean comparison method is utilized for data correction. The feasibility and effectiveness of the proposed abnormal wind power data identification and correction strategies in this study is tested by detailed simulations on the East-China Sea offshore wind power SCADA database.