柔性直流换流站异常监控数据实时预警研究

Real-Time Early Warning Study of Abnormal Monitoring Data in Flexible DC Converter Stations

为了提高对异常数据的预警性能、维护柔性直流换流站的运行稳定性,提出了柔性直流换流站异常监控数据实时预警方法。首先,利用经验规则消除数据碰撞问题,根据数据信息哈希函数的异常值,准确检测出其中的异常数据。然后,利用小波变换对异常监控数据进行预处理,并从异常数据中提取关键变量信息。最后,根据监控过程中出现异常数据的概率,设定预警阈值,再利用预警函数构建实时预警模型。试验结果表明,应用该方法后,误报率被控制在20%以下,平均绝对误差被控制在0.12以下。相比于传统方法,该方法的预警结果更有效、准确率更高。

To improve the early warning performance of abnormal data and maintain the operational stability of the flexible direct current(DC)converter station,a real-time early warning method for abnormal monitoring data in flexible DC converter stations is proposed.Firstly,the empirical rules are utilized to eliminate the data collision problem,and the abnormal data in it are accurately detected according to the abnormal values of the data information Hash function.Then,the wavelet transform is used to preprocess the abnormal monitoring data and extract key variable information from the abnormal data.Finally,according to the probability of abnormal data in the monitoring process,the early warning threshold is set,and then the early warning function is used to construct a real-time warning model.The experimental results show that after applying the method,the false alarm rate is controlled below 20%and the average absolute error is controlled below 0.12.Compared with the traditional method,the early warning results of this method are more effective and the accuracy is higher.