基于红外热成像的变压器故障在线监测系统

发布时间：2018-01-15 15:11

本文关键词：基于红外热成像的变压器故障在线监测系统　出处：《安徽理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：我国经济的快速发展带来了用电需求量的增加,现如今已基本实现电网全覆盖。保障电网安全平稳运行至关重要,在电网中故障多数是由电气设备故障所引起,而变压器又是不可或缺的重要设备,因此变压器的在线故障监测对于保障电网系统运行安全、平稳是必不可少的。红外热成像诊断技术作为一种有效的故障诊断手段,它可以将被监测变压器的热状态分布,快速地进行红外热成像,然后对生成的红外热图像进行分析,分析的过程包括图像的去噪、增强、分割等,从而对变压器可能存在的隐患以及故障做出诊断分析。根据研究课题的实际要求,构建出一种利用红外诊断技术对变压器故障的在线监测系统。首先,将采集的变压器红外图像,分别利用均值法滤波、小波半软阈值滤波法、基于小波半软阈值加均值法滤波去噪,对比三种去噪方法对图像处理的效果,本文选择第三种方法,此方法能有效地抑制噪声,很好地改善图像质量。第二步,对图像用直方图均衡化和小波变换方法,进行增强处理,本文选择能够显著增强对比度的小波变换法。第三步,对图像进行分割处理,将直方图阈值分割法、边缘检测法和Ctsu法自动阈值分割法的仿真结果作对比分析,本文采用基于Otsu法自动阈值分割法进行图像分割。通过这些算法实现变压器红外图像的去噪、增强、分割的过程,为后续故障诊断打下良好的基础。针对故障的识别方法,本文采用基于ASIFT算法对红外图像匹配与识别方法。文中拓扑矩阵修正的方法可以实现变压器故障的自动定位,本文在六种诊断方法中选用了基于相对温差法的变压器故障诊断法,实现了变压器设备红外图像故障的自动诊断。
[Abstract]:The rapid development of economy in our country has brought about an increase in the demand for electricity, and now the total coverage of the power grid has been basically realized. It is very important to ensure the safe and stable operation of the power network, and most of the faults in the power network are caused by the faults of the electrical equipment. Transformer is an indispensable and important equipment, so the on-line fault monitoring of transformer can ensure the safety of power system operation. As an effective means of fault diagnosis, infrared thermal imaging diagnosis technology can monitor the thermal state distribution of transformers and quickly carry out infrared thermal imaging. Then the generated infrared thermal image is analyzed, the analysis process includes image denoising, enhancement, segmentation and so on. In order to make diagnosis and analysis of the potential hidden trouble and fault of transformer, according to the actual requirements of the research project, a kind of on-line monitoring system of transformer fault using infrared diagnosis technology is constructed. First of all. The infrared images of transformer were filtered by means method, wavelet semi-soft threshold filter, wavelet semi-soft threshold plus mean filtering, respectively, and compared the effect of three de-noising methods on image processing. In this paper, we choose the third method, which can effectively suppress the noise and improve the image quality. The second step is to enhance the image using histogram equalization and wavelet transform. In this paper, we choose the wavelet transform method which can significantly enhance the contrast. The third step, the image segmentation, histogram threshold segmentation method. The simulation results of edge detection method and Ctsu automatic threshold segmentation method are compared and analyzed. In this paper, the automatic threshold segmentation method based on Otsu method is used to segment the image. Through these algorithms, the infrared image of transformer is de-noised, enhanced and segmented. It lays a good foundation for subsequent fault diagnosis. In this paper, the method of infrared image matching and recognition based on ASIFT algorithm is adopted. The method of modifying topology matrix can realize the automatic location of transformer fault. In this paper, the transformer fault diagnosis method based on the relative temperature difference method is selected among the six diagnostic methods, and the infrared image fault diagnosis of transformer equipment is realized automatically.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM41;TN219

【相似文献】