基于低质量数据的水电机组变工况状态指标构建方法  

Performance Evaluation of Hydropower Units Under Variable Operation Conditions Based on Low-Quality Data

作　　者：段然[1] 周建中[1] 蔡银辉 王彤 岳林林 DUAN Ran;ZHOU Jian-zhong;CAI Yin-hui;WANG Tong;YUE Lin-lin(School of Civil and Hydraulic Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;Guoneng Daduhe Maintenance and Installation Co.,Ltd.,Leshan 614900,China)

机构地区：[1]华中科技大学土木与水利工程学院,湖北武汉430074 [2]国能大渡河检修安装有限公司,四川乐山614900

出　　处：《水电能源科学》2022年第6期183-187,共5页Water Resources and Power

摘　　要：水电站监测数据与机组运行工况高度关联,且通常存在数据异常和数据缺失等问题。为此,提出了一种基于低质量数据的变工况下水电机组状态指标构建方法。首先采用国内某大型常规水电站的实测水头、有功功率及下机架振动数据构建机组运行数据集,再采用DBSCAN算法清洗异常数据,最后基于GMM算法拟合健康样本的概率密度分布,构建机组健康状态模型。在此基础上,计算待评估样本与健康状态模型之间的负对数似然概率,并将其作为机组性能状态指标。验证分析表明,通过数据清洗能有效识别水电机组运行数据集中的奇异点和离群点,且所构建的健康模型受数据缺失影响较小。The monitoring signal of hydropower station is highly correlated with the operation condition of the unit, and it usually has problems such as data anomaly and data missing. Therefore, a performance evaluation approach of hydropower units under variable operation conditions is proposed based on low-quality data. Firstly, the on-site monitoring data of water head, active power, and vibration of the lower bracket in a large conventional hydropower station in China were adopted to construct the running data set of the unit. Then, the Density Based Spatial Clustering Applications with Noise(DBSCAN) algorithm was used to clean abnormal data in the running data set. Furthermore, the Gaussian Mixture Model(GMM) algorithm was used to fit the probability density distribution of healthy samples to construct the healthy state model. On this basis, the negative log-likelihood probability between the samples to be evaluated and the health state model was calculated as the performance state index. The verification experiments show that the proposed data cleaning method can identify the singularity and outlier in the running data set of hydropower units, and the constructed health state model is less affected by the data loss.

关 键 词：水电机组 低质量数据 变工况 数据清洗 状态评估 

分 类 号：TV734[水利工程—水利水电工程]