摘要
为解决电力纹理图像精准识别率低下的问题,提出基于最优鉴别特征的电力模糊图像边缘纹理智能识别方法。在电力模糊图像的最优特征鉴别子集中,通过计算提取复杂度的方式,统计图像纹理的邻类参量,完成基于最优鉴别特征的电力模糊图像边缘纹理参量提取。在此基础上,利用边缘神经网络中电力图像节点的分布情况,计算智能平滑参数,并根据现有模糊图像的具体数量,对识别流程进行完善创新,实现新型智能识别方法的搭建。与现有识别手段相比,应用基于最优鉴别特征的电力模糊图像边缘纹理智能识别方法后,横波、纵波电力纹理图像识别准确率的最大值均超过90%,精准识别率低下的问题得到有效解决。
To solve the problem of low precision recognition rate of electric power texture image,an intelligent recognition method of electric power fuzzy image edge texture based on optimal discriminant features is proposed.In the optimal feature discriminant subset of power blurred image,the edge texture parameters of power blurred image based on the optimal discriminant feature are extracted by calculating the extraction complexity and counting the neighborhood parameters of image texture.On this basis,the distribution of power image nodes in the edge neural network is used to calculate intelligent smoothing parameters,and according to the specific number of existing blurred images,the recognition process is improved and innovated to achieve the construction of a new intelligent recognition method.Compared with the existing recognition methods,the maximum recognition accuracy of S-wave and P-wave power texture images is more than 90% after applying the intelligent recognition method of edge texture of power fuzzy images based on the optimal discriminant features,which effectively solves the problem of low precision recognition rate.
作者
张陵
常喜强
高宝琪
王学民
王志远
赵建平
ZHANG Ling;CHANG Xiqiang;GAO Baoqi;WANG Xuemin;WANG Zhiyuan;ZHAO Jianping(State Grid Xinjiang Electric Power Co.,Ltd.,Urumqi 830001,China;State Network Urumqi Power Supply Company,Urumqi 830001,China)
出处
《自动化与仪器仪表》
2019年第11期60-63,共4页
Automation & Instrumentation
基金
国家自然科学青年基金(No.61581051)
关键词
鉴别特征
电力图像
智能识别
最优子集
邻类参量
神经网络
平滑参数
discriminant features
power image
intelligent recognition
optimal subset
neighborhood parameters
neural network
smoothing parameters