随着我国智能电网建设进程的推进, 其中的智能电气设备能够自动识别故障显得尤为重要, 许多电气设备故障都伴有过热现像并具有区域性的特点, 体现在红外图像温度与其灰度值具有非线性的映射关系。针对电气设备红外图像对比度差、细节不明显等特点, 提出了一种基于非线性 NSCT(Nonsubsampled Contourlet Transform)变换的图像增强算法, 在算法中构造非线性增强匹配函数, 能够对图像强弱边缘进行不同程度的增强, 并对噪声有一定的抑制作用。对红外图像进行增强后通过拓扑矩阵修改, 实现了图像较高灰度值区域的识别标记, 从而实现了电气设备温度过高区域的自动定位, 之后采用相对温差法对设备是否为故障进行诊断。实验结果表明, 本文方法能够迅速有效地对电气设备疑似过热故障进行自动诊断和定位。

With the advancement of the smart grid construction in China, it becomes particularly important for the smart electrical equipment to automatically identify the fault. Many electrical equipment failures are associated with overheating and have regional characteristics. As a result, there is a nonlinear mapping relationship between infrared image temperature and grey value. Being aimed at poor contrast and unconspicuous details of electrical equipment infrared image, an image enhancement algorithm based on NSCT(nonsubsampled contourlet transform)is proposed and nonlinear enhancement matching function is constructed. By this algorithm, the edge of the image is enhanced in varying degrees and the noise of the image is controlled to a certain extent. After topology matrix being modified, higher image gray value area is identified and electrical equipment’s high temperature area is located automatically. With the relative temperature difference algorithm, electrical equipment failure is diagnosed. Experiment results show that the method can locate and identify electrical equipment suspected overheating fault quickly and effectively.