摘要
针对电气设备运行状态图像的特点,提出将支持向量机(SVM)分类器应用于多种电气设备运行状态识别中。首先利用C-均值聚类法,分割出运行状态指示牌的汉字或数字部分;再利用K-L变换提取出运行状态的特征向量;最后利用支持向量机分类方法进行状态识别。试验结果表明:支持向量机分类方法对于小样本情况,具有良好的分类能力,适合多种电气设备运行状态的分类,并能获得比神经网络方法更好的识别性能。不同的分类核函数的相互比较分析表明,Sigmoid核函数最适合电气设备运行状态的分类识别。
According to the characters of operation condition image of electricity equipments, a method of the recognition of electricity equipments operation condition was put up based on Support Vector Machine （SVM）. First chinese character or number operation condition images of electricity equipments were segmented with color segmentation.Then, characteristic vector of circulation state image of electricity equipments was extracted using K-L transform.At last, classification method of SVM for state recog- nition was used. Experimental results showed that classification method of SVM had better classification ability for smaller sam- ples situation, which adapts to classification of many electricity equipments operating condition, and could get better recognition result than that of neural networks. Comparing with all the kernel functions, kernel function of Sigmoid was the best way to recognition of electricity equipments operation condition.
出处
《沈阳农业大学学报》
CAS
CSCD
北大核心
2008年第5期638-640,共3页
Journal of Shenyang Agricultural University
基金
辽宁省自然科学基金(20042102)
关键词
支持向量机
电气设备运行状态
图像识别
C-均值聚类法
support vector machine （SVM）
operation condition of electricity equipments
image recognition
C-mean clustering
