变压器故障诊断与定位研究

发布时间：2019-01-27 17:49

【摘要】：近年来,随着电力工业的发展,电力系统装机容量日益增长,变压器故障时有发生,此时,及时的发现并处理变压器的故障显得格外的重要。在样本数量较小的情况下,支持向量机(SVVM)依然可以解决非线性、高维等问题,根据此优势,本文建立了基于支持向量机的变压器故障诊断模型和故障定位模型。故障诊断的结果主要是由支持向量机的参数所决定的,所以,本文还提出用布谷鸟算法(CS)来优化支持向量机的参数。通过仿真实验表明,相比于其他算法,该算法能够提高变压器故障诊断以及故障定位的正确率。本文的主要研究内容如下:(1)在变压器故障诊断模型中,利用布谷鸟算法优化支持向量机的惩罚参数C和核参数g,为了提高布谷鸟算法的优化能力,引入一种新的惯性权重ω,并将改进的布谷鸟算法(WCS)与最速下降法(SD)相结合,得到新算法(SDWCS)。利用该算法进行SVM参数的寻优,克服了基本的SVM模型容易陷入局部最优的缺陷。(2)在变压器故障诊断模型中,建立了基于支持向量机的分类模型,采用SDWCS算法优化支持向量机参数,利用LibSVM工具箱在MATLAB软件平台上训练支持向量机,并通过实例仿真,将该故障诊断模型与布谷鸟算法(CS)、粒子群算法(PSO)、遗传算法(GA)、网格搜索算法(GS)相比较。(3)在变压器故障定位模型中,将变压器的油色谱信息和电气试验特征结合,建立基于二叉树的支持向量机多层分类模型,采用SDWCS算法优化支持向量机参数,并逐步对变压器故障进行定位。通过实例证明,该故障定位模型能够快速的判断出变压器发生故障的部位,而且定位的准确率也有所提高,以做到快速的处理故障。
[Abstract]:In recent years, with the development of power industry, the installed capacity of power system is increasing day by day, and transformer faults occur frequently. At this time, it is very important to find and deal with transformer faults in time. In the case of small number of samples, SVM (SVVM) can still solve nonlinear and high-dimensional problems. Based on this advantage, a transformer fault diagnosis model and fault location model based on SVM are established in this paper. The result of fault diagnosis is mainly determined by the parameters of support vector machine (SVM). Therefore, this paper also proposes to optimize the parameters of support vector machine by Cuckoo algorithm (CS). The simulation results show that compared with other algorithms, the proposed algorithm can improve the accuracy of transformer fault diagnosis and fault location. The main contents of this paper are as follows: (1) in the transformer fault diagnosis model, Cuckoo algorithm is used to optimize the penalty parameter C and kernel parameter g of support vector machine, in order to improve the optimization ability of cuckoo algorithm. A new inertial weight 蠅 is introduced, and the improved cuckoo algorithm (WCS) is combined with the steepest descent method (SD) to obtain the new algorithm (SDWCS). The algorithm is used to optimize the SVM parameters, which overcomes the defect that the basic SVM model is prone to fall into local optimum. (2) in the transformer fault diagnosis model, a classification model based on support vector machine is established. SDWCS algorithm is used to optimize the parameters of support vector machine, and LibSVM toolbox is used to train support vector machine on MATLAB software platform. The fault diagnosis model and (CS), particle swarm optimization algorithm (PSO), of cuckoo algorithm are simulated by an example. The genetic algorithm (GA), grid search algorithm (GS) is compared. (3) in the fault location model of transformer, the multi-layer classification model of support vector machine based on binary tree is established by combining the oil chromatographic information of transformer with the characteristics of electrical test. The parameters of support vector machine are optimized by SDWCS algorithm, and transformer fault location is carried out step by step. The example shows that the fault location model can quickly determine the fault location of transformer, and the accuracy of location is also improved, in order to deal with the fault quickly.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM407

【相似文献】