针对配电变压器的数据清洗方法

发布时间：2018-11-06 11:40

【摘要】：配电变压器作为配用电系统的关键环节,由于分布面广、网络结构复杂、运行环境恶劣等客观条件的制约,其数据在实际运行中常存在大量异常和缺失情况。为解决上述问题,以现实中的配电变压器数据为研究样本,首先分析了配电变压器数据异常和数据丢失2类情况的产生原因,之后采用小样本数据进行算法研究,针对配电变压器数据异常问题提出基于Kernel Smoothing技术的异常数据识别方法,针对配电变压器数据丢失问题提出了基于Pearson相关系数与回归模型相结合的缺失数据恢复方法。为验证所提方法能够满足TB级数据的运算效率,采用6台服务器搭建Spark并行计算结构,对大样本的配电变压器数据进行了异常数据的识别和缺失数据的恢复,相关结果证明了所提方法的实际价值。
[Abstract]:As the key link of distribution power system, due to the restriction of objective conditions, such as wide distribution, complex network structure and poor operating environment, there are often a large number of anomalies and deficiencies in the actual operation of the distribution transformer. In order to solve the above problem, taking the distribution transformer data in reality as the research sample, this paper first analyzes the causes of the abnormal data and the data loss of the distribution transformer, and then uses the small sample data to carry on the algorithm research. In order to solve the problem of abnormal data of distribution transformer, a method of identifying abnormal data based on Kernel Smoothing technology is put forward, and a method of recovering missing data based on Pearson correlation coefficient and regression model is proposed to solve the problem of data loss of distribution transformer. In order to verify that the proposed method can meet the operational efficiency of TB level data, six servers are used to construct the Spark parallel computing structure, and the abnormal data and the missing data are identified and recovered from the large sample distribution transformer data. The results show that the proposed method is of practical value.
【作者单位】：华北电力大学电气与电子工程学院;中国电力科学研究院配电研究所;
【分类号】：TM421

【相似文献】