海上风电结构安全监测数据异常识别

Anomaly Identification of Structural Safety Monitoring Data for Offshore Wind Power Projects

海上风电结构安全监测数据的可靠性和完整性是分析海上建筑物结构安全,评估其运行状态的基础,也是开发预警模型的前提。对原始监测数据进行异常值识别和过滤,提高数据有效性,具有重要的意义。通过比较3种基于距离和密度的算法在监测数据异常识别中的准确率和误差,发现DBSCAN聚类算法在识别海上风电结构安全监测数据的异常值中表现更优。剔除异常值后的监测数据质量较好,既能反映正常观测期间的变化规律,也可为后续开发预警模型提供数据支撑。

The reliability and integrity of the structural safety monitoring data of offshore wind power projects are the basis for the structural safety analysis and operation status assessment of offshore structures,and also the prerequisite for the development of early warning models.It is of great significance to identify and filter the outliers from the original monitoring data to improve the data validity.The accuracy and error of three algorithms based on distance and density in anomaly identification of monitoring data are compared.It is found that DBSCAN clustering algorithm is of the best performance in anomaly identification of structural safety monitoring data for offshore wind power projects.The monitoring data after eliminating the anomaly data is of better quality,which can not only reflect the data variation during the normal observation,but also provide data support for the subsequent development of early warning models.