基于多域联合特征的密封继电器多余物材质识别  

Material indentification of redundant material of sealed relays based on multi-domain joint feature

作　　者：高亚杰 蒋爱平[1] 王国涛[1] 孙志刚 GAO Yajie;JIANG Aiping;WANG Guotao;SUN Zhigang(College of Electronic Engineering,Heilongjiang University,Harbin 150080,China)

机构地区：[1]黑龙江大学电子工程学院,哈尔滨150080

出　　处：《黑龙江大学自然科学学报》2021年第3期335-341,共7页Journal of Natural Science of Heilongjiang University

基　　金：国家自然科学基金资助项目(51607059);黑龙江省自然科学基金资助项目(JJ2020LH1310,QC2017059);黑龙江省博士后基金资助项目(LBH-Z16169);黑龙江省高校基本科研业务费(2020-KYYWF-1006);黑龙江省教育厅科技成果培育项目(TSTAU-C2018016)。

摘　　要：提出了一种基于多域联合特征的密封继电器多余物材质识别方法,利用声音转图像技术,将一维的多余物声音信号转换为二维的图像信号,从多余物信号生成的声谱图中提取纹理特征。结合时域和频域提取能量密度、左右对称程度和波峰系数等特征组成联合特征向量,采用随机森林算法进行模型训练和分类识别。结果表明,本方法对金属和非金属的识别准确度可提高至90%,具体材质的综合识别准确度达到78%,高于目前检测系统的准确率。A sealed relays redundant material identify method based on multi-domain joint features is proposed.The signal processing method of sound-to-image is adopted to convert one-dimensional redundant sound signals into two-dimensional image signals,and the texture features are extracted from the acoustic spectrum generated by redundant material signals.Combined with the time domain and frequency domain,energy density,the left-right symmetry degree,the crest coefficient and other features are extracted to constitute a joint feature vector,and the random forest algorithm is used for model training and classification recognition.The experimental results show that the recognition accuracy of the proposed method for metal and non-metal can be increased to 90%,and the comprehensive recognition accuracy for specific materials can research 78%,which is higher than the accuracy of the current detection system.

关 键 词：密封继电器 多余物材质识别 声谱图 联合特征向量 

分 类 号：Q939.97[生物学—微生物学]