基于DBSCAN的风机功率异常数据清洗

Cleaning of Abnormal Data of Wind Turbine Power Based on DBSCAN

摘要: 风功率曲线是评价风电机组性能的重要指标，对风电场整体的运行管理具有重要意义。在风电机组实际运行过程中，由于设备故障及自然因素等原因的影响会导致数据采集与监视控制系统(SCADA)采集的数据中存在大量异常数据，导致风功率曲线评价不准确。本文从异常数据的产生机理分析，将数据分为0功率堆积数据、恒功率限电数据和分散型异常数据，根据不同类型数据特征，提出了基于DBSCAN和区间DBSCAN (DBSCAN-Interval DBSCAN)组合的异常检测模型，实现了对运行数据的清洗。最后，将本方法应用到某风场全年的风机采集数据中，对其进行数据清洗，结果表明该方法可以有效地检测和分离运行数据中的异常数据，在保证数据完整性的基础上提高了数据质量，显著提高了风电机组性能分析的准确性。

Abstract: The wind power curve is an important indicator for evaluating the performance of wind turbines, and is of great significance to the overall operation and management of the wind farm. However, due to equipment failures and natural factors, there will be a large number of abnormal data in the data collected by the data collection and monitoring control system (SCADA) in practices, making the wind power curve inaccurate. In this paper, aimed to clean the data of wind power curve, the data is divided into zero power accumulation data, constant power limit data and scattered abnormal data based on the analysis of the generation mechanism of abnormal data firstly. Then, according to the characteristics of the different types of data, a combination of DBSCAN and interval DBSCAN (DBSCAN-Interval DBSCAN) method is proposed, which realizes the anomaly operating data detection and cleaning. Finally, this method is applied to the annual wind turbine collection data of a wind farm to clean the data. The results show that DBSCAN-Interval DBSCAN method can effectively detect and clean abnormal data in the operating data, enforce the data integrity, and improve data quality, which significantly enhanced the accuracy of performance analysis of wind turbines.