【作者】 毕强；

【导师】 孙才新；

【作者基本信息】 重庆大学 ， 电气工程， 2002， 硕士

【摘要】 变压器作为电力系统的枢纽设备，其运行可靠性直接关系到电力系统的安全与稳定。本文针对应用油中溶解气体分析方法进行变压器绝缘监督时所遇到的主要技术难点，提出了提高变压器故障诊断的准确性、可靠性的几种模诊断方法，主要研究内容有： 通过对判断变压器故障常用的三比值法和改良电协研法的深入分析，其诊断准确率较高，但对故障原因、故障现象和故障机理间同时存在不确定性和模糊性的变压器等电气设备的故障诊断，难于满足工程应用的需要；在处理变压器等结构复杂设备的多症状、多原因故障诊断时，模糊关系矩阵可以全面反映这种症状与故障类型间的因果关系，进而提出了变压器故障诊断的模糊综合诊断模型；同时作者还深入分析了模糊算子的特性；针对常用的模糊评判结果的模糊集的集化方法的不足性，提出了将模糊综合诊断与模糊规则推理结合起来进行故障诊断的方法，能达到较好的效果。即首先采用模糊推理对故障原因进行“过滤”，滤掉可能性极小的原因，然后再利用规则推理对剩下的可能原因进行验证，得出最终结论。 根据以变压器DGA数据为特征量的样本空间各样本差异特性以及样本在空间R~s的分布特性，首次提出了基于势函数自适应加权的变压器绝缘故障诊断的模糊c-均值聚类模型；同时，从s维样本空间的F~c-划分几何特性出发，提出了一种求取样本集的类势有效邻域半径和自适应求取聚类数和聚类中心初值的方法；对一个待诊断样本，设计了基于类势密度函数意义下的属性测度和诊断准则。 经过大量的诊断实例表明：本文的模糊综合诊断与模糊规则推理结合起来的故障诊断方法对电力研究变压器绝缘故障诊断是有效的，能大大减少错判、误判比率；自适应加权模糊聚类方法能对变压器的故障样本正确分类，同时能有效判别故障类型，诊断的准确性大大高于常用的三比值法。更多还原

【Abstract】 The operation reliability of the power transformer,which is the major equipment in power system,directly related to the safety and stability of whole power system. In accordance with the technological difficulties encountered in the process of insulation supervision based on the Dissolved Gases Analysis (DGA),several kinds of model and method are presented to improve the reliability and precision of fault diagnosis of the power transformer. Main research content includes:By deeply studying the common transformer faults diagnosing methods,such as three-ratio methods and improved electrical committee agreements,several shortcomings such as uncertainness judgment when the fault reasons,phenomenon and principles come out together while can not consistent to each other etc. For this reason,the old methods can not fully meet the need to engineering practical application.Considering fuzzy relationship matrix can fully represents the causality between fault symptoms and fault types,when diagnosing complex equipments with multiple symptoms and fault causes such as power transformer,a synthetic fuzzy diagnosing model is firstly proposed to diagnose transformer’s insulation faults based on DGA in this paper. Meanwhile,by further study the character of fuzzy factors,aimed at correcting the defects in classifying fuzzy sets of common fuzzy judgment,a method are put forward by unifying fuzzy synthetic diagnosis and fuzzy principle reasoning. Firstly,fuzzy principle reasoning is applied to filter fault reasons with low possibility,then fuzzy synthetic diagnosis is utilized to test all the left fault reasons and pick up the reasonable ones.In connection with the difference and distribution characteristic of the samples in sample space RS based on DGA,a new self-adapted weight fuzzy omean clustering model of fault diagnosis of the power transformer based on the potential function is proposed. Meanwhile,from the aspect of geometry characteristic of FC-divided in s dimension sample space,a method is proposed for the purpose of getting an effective adjacent radius,adaptive cluster number c and original cluster center of X sample set. For the diagnosissample x,the property measure and diagnosis rule are proposed,which under the condition of potential density function that determine c number of optimal fuzzy cluster P1.Plenty of diagnosis examples demonstrate the validity of the gray relationship analysis method in diagnosing power system insulation faults. Not only the misjudgment rate can be reduced,but also the transformer fault incidences can be classified by using self-adapted weight fuzzy clustering method. Further more,the transformer insulation condition can be analyzed,as well as the fault position can be located by recognizing the fault types. The diagnosing precision is much higher comparing with former three-ratio method.更多还原