基于聚类主成分分析法的变压器参数关联度评估

为了有效管理和监测电力变压器的健康状态,在对变压器正常运行的海量数据进行分析的基础上,构建变压器参数关联度评估方法。首先,为了对变压器运行参数进行参数分类定性,基于系统聚类的方法构建变压器参数样本分类流程算法,并设计了相应的流程图;然后,通过引入聚类以及主成分分析方法,建立基于聚类的主成分分析的变压器主要影响参数关联度评估模型;最后通过潼南电力局实际运行数据验证模型的可靠性和有效性,准确分析了变压器运行参数之间的定量关系,可以为变压器故障预测研究提供重要依据。

Correlation Degree Evaluation of Transformer Parameters Based on Clustering and Principal Component Analysis

To manage and monitor effectively the health status of power transformer,an evaluation method of correlation degree between transformer parameters is constructed by in-depth analysis of massive amounts of data in power transformer. Firstly,a clustering algorithm is constructed based on system cluster method to finish the qualitative class of running parameters in power transformer,and then the flow chart is designed correspondingly. Then,an evaluation model of correlation degree among primary transformer parameters is established by combining with the clustering and principal component. Finally,using the real running data provided by Chongqing Tongnan Power Supply co.,LTD verifies the reliability and effectiveness of the model and algorithm. The result shows that the quantitative relation among running parameters of power transformer may be recognized exactly,so it can provides important support for the fault diagnosis and predict of power transformer.