基于红外图像处理的变电设备识别与热故障诊断

发布时间：2018-07-07 21:26

本文选题：红外图像 + 图像处理　；参考：《上海电机学院》2017年硕士论文

【摘要】：变电设备热故障诊断对电网事故的预防起着非常关键的作用。红外检测技术已经广泛应用于变电设备的状态监测与热故障诊断,但是目前国内仍普遍采用人员手持红外热像仪的方式进行定期巡检,效率较低,不能及时发现热故障。本文对变电设备红外图像识别所必需的图像去噪、图像分割、特征提取和分类识别方法,以及变电设备热故障红外诊断方法进行研究,并将研究成果应用于变电设备红外图像识别与热故障诊断仿真系统,为变电设备热故障实时诊断的实现奠定基础。首先,通过分析红外图像的特点及常见噪声,针对传统小波阈值去噪的不足进行了改进,并提出中值滤波加改进的小波阈值去噪的方法来去除变电设备红外图像中的混合噪声,通过实验验证了所提出方法的有效性。其次,根据变电设备红外图像的特点提出了改进的区域生长方法,分割效果比传统的区域生长法更好,并结合使用边缘检测和数学形态学方法分割出目标设备的边缘。接着,针对变电设备红外图像获取过程中可能会出现的图像旋转、缩放和平移现象,采用Hu不变矩对图像中变电设备的形状特征进行提取,然后使用支持向量机(SVM)进行变电设备红外图像的识别,识别的准确率较高。最后,根据红外测温技术,使用基于相对温差法的热故障诊断方法,通过设计仿真系统,实现了变电设备红外图像识别与热故障的诊断。
[Abstract]:Thermal fault diagnosis of substation equipment plays a key role in the prevention of power network accidents. Infrared detection technology has been widely used in the condition monitoring and thermal fault diagnosis of substation equipment, but at present, it is still widely used in our country to carry out regular inspection by holding infrared thermal imager. The efficiency is low, and thermal fault can not be found in time. In this paper, the necessary methods of image denoising, image segmentation, feature extraction and classification recognition for infrared image recognition of substation equipment are studied, as well as infrared diagnosis method for thermal fault of transformer equipment. The research results are applied to the infrared image recognition and thermal fault diagnosis simulation system of substation equipment, which lays a foundation for the realization of real-time thermal fault diagnosis of substation equipment. Firstly, by analyzing the characteristics of infrared image and common noise, the traditional wavelet threshold de-noising method is improved, and the median filter and improved wavelet threshold de-noising method is proposed to remove the mixed noise in the infrared image of transformer equipment. The effectiveness of the proposed method is verified by experiments. Secondly, according to the characteristics of infrared image of substation equipment, an improved region growth method is proposed. The segmentation effect is better than the traditional region growth method, and the edge of the target equipment is segmented by using edge detection and mathematical morphology. Then, aiming at the phenomenon of image rotation, scaling and translation, Hu invariant moment is used to extract the shape feature of the transformer device in the image, which may appear in the process of infrared image acquisition. Then support vector machine (SVM) is used to recognize infrared image of transformer equipment, and the recognition accuracy is high. Finally, according to the infrared temperature measurement technology, the thermal fault diagnosis method based on the relative temperature difference method is used, and the infrared image recognition and thermal fault diagnosis of the transformer equipment are realized by designing the simulation system.
【学位授予单位】：上海电机学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TM507

【参考文献】