Offshore wind turbine wind speed power anomaly data cleaning method based on RANSAC regression and DBSCAN clustering

As the transition to renewable energy sources intensifies, offshore wind power has rapidly emerged as a critical component of the new energy landscape. However, the complexity of offshore turbine failure mechanisms, the difficulty of operation and maintenance, and the high cost all put forward strict requirements for monitoring and real-time health analysis of wind turbines. Operational data from wind turbine generators often includes numerous anomalies such as noise, faults, and instances of power abandonment, which reduce the data’s availability and consequently, the precision of wind power forecasts, fault diagnostics, and other research endeavors. This study scrutinizes the distribution characteristics of such anomalous data within wind power curves and introduces a data-cleaning methodology that leverages a fusion algorithm combining the Random Sample Consensus (RANSAC) and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithms. A comparative validation assesses the algorithm’s effectiveness in cleaning, efficiency, and the judiciousness of data elimination. The findings indicate that the proposed method adeptly identifies the scope of anomalous data, offering substantial value for engineering applications.