基于Prophet-GMM的大坝监测数据异常检测算法

Anomaly Detection Algorithm of Dam Monitoring Data Based on Prophet⁃GMM

大坝监测数据受环境等因素影响,往往存在异常数据,异常数据的检测对于大坝的正常运行起着不可或缺的作用,但是传统异常检测算法对于大坝监测数据往往达不到精度要求。提出了一种基于Prophet-GMM的异常检测算法,利用Prophet算法较好的拟合性能对大坝数据进行拟合,由拟合数据与实测数据求残差序列,再利用GMM算法对残差序列进行聚类,从而准确识别出异常值。结果表明:Prophet-GMM法对于不同类型的大坝监测数据都能准确识别出异常值,与传统检测算法相比,在查准率、查全率及准确率3个检测指标上,均有较为明显的提升。

Due to the influence of environment and other factors,there are often abnormal data in dam monitoring data and the detection of abnormal data plays an indispensable role in the normal operation of the dam.However,the accuracy of traditional anomaly detection algo⁃rithms for dam monitoring data often fails to meet the requirements.In this paper,an anomaly detection algorithm based on Prophet GMM was proposed.The better fitting performance of Prophet algorithm was used to fit the dam data and the residual sequence was obtained from the fit⁃ting data and the actual data.Then,the residual sequence was clustered by GMM algorithm to accurately identify the abnormal value.The test results show that the method proposed in this paper can accurately identify outliers for different types of dam monitoring data.Compared with the traditional detection algorithm,it has significantly improved the detection indicators of precision,recall and accuracy.