基于特征降维和神经网络的电能表内异物声音自动识别Automatic Recognition of Foreign Object Sound in the Electricity Meters Based on Feature Dimension Reduction and Neural Network
张进;吴健;欧习洋;欧熙;
摘要(Abstract)：
电能表是国家强制检定的电能计量工具,其计量的精确性影响着千家万户的利益。传统的人工检测方式不仅效率低而且检测结果不稳定。随着声学检测技术的日趋成熟,采用声学检测的方式来检测电能表内的异物已成为实现工厂自动化的大势所趋。针对现有半自动的人工检测电能表异物方式,提出一种基于特征降维和神经网络的电能表内的异物声音自动识别方法。该方法充分利用声音的时、频域特征系数和倒谱系数,先对声音信号进行通道转换、预处理和数字降噪,再对声音信号进行时、频域和倒谱分析,并同时提取其短时特征系数及改进后MFCC系数。将声音特征通过PCA降维后输入基于Adaboost算法聚类后BP神经网络分类识别,并与传统的BP神经网络分类进行比较,证明了该方法的有效性。这里给出了电能表异物自动识别技术实现的具体步骤,并通过MATLAB仿真实验证明了该方法的有效性,BP神经网络的平均识别率较高,可达到95%以上,并且计算复杂度小易于实现。
关键词(KeyWords)： 电能表异物检测;改进MFCC;PCA特征降维;Adaboost聚类;BP神经网络
基金项目(Foundation): 基于神经网络算法的电能表异物自动检测技术研究（2017渝电科技10#）;; 电能表异物自动检测技术优化应用研究（FDRS05HT1707002MIG）
作者(Author): 张进;吴健;欧习洋;欧熙;
Email:
DOI: 10.19356/j.cnki.1001-3997.2021.03.053
参考文献(References)：
- [1]戚龙，赵丹.基于BP神经网络的非特定人语音识别算法[J].科学技术与工程，2017,17(31):277-282.(Qi Long,Zhao Dan.Unspecified human speech recognition algorithm based on BP neural network[J].Science Technology and Engineering,2017,17(31):277-282.）
- [2]张红伟.基于深度神经网络的蒙古语语音识别系统声学模型的研究[D].内蒙古：内蒙古大学，2017.(Zhang Hong Wei.Research on acoustic model of mongolian speech recognition system based on deep neural network[D].Inner Mongolia:Inner Mongolia University,2017.）
- [3]白琳，黄梓瑜，叶程.基于BP神经网络的车辆声音信号识别[J].自动化技术与应用，2014,33(2):64-66+86.(Bai Lin,Huang Zi-yu,Ye Cheng.Vehicle sound signal recognition basedon BPneuralnetwork[J].Automation Technologyand Application,2014,33(2):64-66+86.）
- [4]赵桂清.BP神经网络信息融合的汽车载重测量方案[J].机械设计与制造，2017(12):134-137.(Zhao Gui-qing.Automobile load measurement scheme based on BP neural network information fusion[J].Machinery Design&Manufacture,2017(12):134-137.）
- [5]赵亮，彭琳.基于GA-BP神经网络的发动机失火诊断[J].机械设计与制造，2017(10):117-120+125.(Zhao Liang,Peng Lin.Enginemisfirediagnosisbasedon GA-BPneuralnetwork[J].Machinery Design&Manufacture,2017(10):117-120+125.）
- [6]Chang Chuan-yu,Chang Yi-ping.Application of abnormal sound recognition system for indoor environment,2013 9th International Conference on Information,Communications&Signal Processing,Tainan,2013:1-5.
- [7]L.Li.Classificationofheartsoundsignalswith BPneuralnetworkand logistic regression,2017 Chinese Automation Congress(CAC),Jinan,2017:7380-7383.
- [8]Y.Li,X.Yang,Y.Sun,C.Hu and T.Zeng,Robust sound recognition method based on human auditory bionic processing[C].IET International Radar Conference,Hangzhou,2015:1-5.
- [9]J.S.Jaffe,P.L.D.Roberts and P.Roux,"Using underwater ambient sound to localize swarms of underwater vehicles,"OCEANS 2015-Genova,Genoa,2015:1-4.
- [10]A.Moukadem,A.Dieterlen and C.Brandt,Shannon Entropy based on the S-Transform Spectrogram applied on the classification of heart sounds,2013 IEEE International Conference on Acoustics,Speech and Signal Processing,Vancouver,BC,2013:704-708.
- [11]S.Sameh,Z.Lachiri,Reassignment method for audio sounds classification,2016 2nd International Conference on Advanced Technologies for Signal and Image Processing(ATSIP),Monastir,2016:638-642.