电力变压器故障的客观熵权识别及诊断方法

摘要

为了有效管理和监测电力变压器的健康状态,在对变压器油中溶解气体数据进行分析的基础上,建立了一种基于客观熵权的电力变压器故障信息模式识别及诊断模型.首先,在定义包含电力变压器故障模式全局信息的矩阵范式基础上,引入信息熵权理论构建故障特征信息的客观熵权精确量化模型;然后,基于距离和投影原则构建了故障模式判别准则函数,并通过准则函数对模式进行排序,运用综合排序结果进行故障测试模式分类,得到用于判断故障类型的基准类心向量;最后,运用基于类心欧氏距离的方式判别故障测试样本所属的类别,实现变压器故障的客观熵权识别及诊断.利用从某电力公司采集到的120组电力变压器油中溶解气体样本进行实例验证,结果表明,所提出的方法能克服传统的三比值故障诊断方法存在无编码以及边界编码模糊致误判的问题.%In order to effectively manage and monitor the running status of power transformer, on the basis of analyzing transformer data at running time, a transformer fault information pattern recognition and diagnosis model using the objective entropy weight method is developed.First of all, in conjunction with the information pattern recognition theory and information entropy and other related theories, the way of defining the fault mode matrix data of all the information is used to build the fault mode matrix.At the same time, the information entropy is used to represent the precise quantitative fault information.Next, fault pattern criterion functions are created on the principles of distance and projection.Fault training patterns are sequenced by the above criterion functions, and then the sequencing patterns are utilized to classify fault test patterns.Finally, by calculating the quasi-core vector using the classification results, the test pattern could be classified by Euclidean distance criterion.To validate the reliability and validity of the established model, one hundred and twenty samples of oil gas provided by an electric power company are adopted to test this model.Compared with the three ratios of the traditional fault diagnosis method, the proposed method has overcome the disadvantages of no coding and coding fuzzy boundary which would lead to misjudging problem in the traditional three-ratio method.