频繁模式的水电信号异常检测

Anomaly detection of hydropower signals in frequent pattern

水电站监控系统信号监测是运行值班的核心工作,研发水电站系统信号的异常检测方法对保障水电站安全稳定运行具有重要意义。然而,水电站系统信号具有信息量大、信号关系复杂等特征,使得信号异常检测具有挑战性。目前关于水电信号异常检测的工作较少,且已有研究也仅基于简单的历史状态或者关联关系构建检测方法。文中认为系统信号属于高频流式数据,且前后信号之间存在相关性,监控过程中考虑信号之间的相关性关系比关联性更好,故创新地提出信号之间的三种关系:原因关系、伴生关系和共生关系,并基于这三种关系设计一种水电信号异常检测方法。通过瀑布沟水电机组的简报历史数据进行实验,结果表明所提方法的召回率可以达到86.92%,验证了检测方法的有效性。

The signal monitoring of hydropower station monitoring system is the core work of operation duty.The research and development of abnormal detection methods for hydropower station system signals is of great significance for ensuring the safe and stable operation of hydropower stations.However,the signal of hydropower station system has the characteristics of large amount of information and complex signal correlation,which makes signal anomaly detection challenging.Currently,there is little work on hydropower signal anomaly detection.In the existing research,the detection methods are built on the basis of simple historical states or correlation relationships.In this paper,it is considered that the system signal belongs to high⁃frequency flow data,and there is correlation between the front and rear signals.In the monitoring process,it is better to consider the correlation relationship between the signals rather than the relevancy relationship.Therefore,three relationships among the signals are innovatively proposed:causative relationship,associated relationship,and symbiotic relationship.Based on these three relationships,a hydropower signal abnormality detection method is designed.The experimental results based on the brief historical data of the Pubugou hydropower unit show that the recall rate of the proposed method can reach 86.92%,verifying the effectiveness of the detection method.