摘要
借助计算机视觉替代人工进行巡视,通过图像识别获取的电网动态参数与门限值进行比较是评判电网运行状态的一种新途径。其中仪表图像的自动识别是获取动态参数的关键环节,该文提取了图像中颗粒目标的长度比、紧密性和简单度3个特征不变量,应用RBFNN实现了表盘关键元素的自动分类。通过对指针式仪表图像的识别实验,证明了输入RBFNN的特征不变量在仪表元素识别中是稳定的,对噪声不敏感,引入图像识别技术可大大优化电力系统运行状态的监测过程。
Acquiring the dynamic parameters based on computer vision instead of manual patrol, and via comparing power meter reading with its preliminary definition threshold, which is a new automatic monitoring scheme of power system running. In the process of power meter automatic recognition, three particle feature invariants, include length ratio, compactness and simplicity factor have been extracted. The RBFNN is utilized in the dial plate elements recognition. The meter identifying experiments proved that the RBFNN input invariants are stable and proper, and it is insensitive to the background noise. It can optimize the power system monitoring with the image-analyzing introducing.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2006年第10期104-108,共5页
Proceedings of the CSEE
关键词
动态参数
计算机视觉
特征不变量
径向基函数神经网络
表盘元素识别
dynamic parameter
computer vision
feature invariants
radial basis functions neural network
dial plate elements recognition