基于信号分析的发电机组辐射噪声盲源分离和识别  

Blind Source Separation and Identification of Generator Radiation Noise Based on Signal Analysis

作　　者：景亚兵[1,2] 刘昌文[1] 毕凤荣[1] 

机构地区：[1]天津大学内燃机燃烧学国家重点实验室,天津300072 [2]天津大学内燃机研究所,天津300072

出　　处：《内燃机工程》2017年第2期141-145,共5页Chinese Internal Combustion Engine Engineering

基　　金：国家科技支撑计划(2015BAF07B04)

摘　　要：具有独立分量的经验模态分解算法结合了独立成分分析算法和经验模态分解算法的优点,可通过单个观测信号将噪声源分解成一系列独立分量,保证了分离出的声源的独立性,同时利用小波变换技术对各个独立分量进行了分析,通过时频分析结果确定了各独立分量与发电机组各噪声源的对应关系,并通过试验进行了验证。研究结果表明:这些独立分量分别对应着配气机构噪声、平衡轴驱动齿轮噪声、正时齿轮噪声和进气噪声。For generator radiation noise, the algorithm for empirical mode decomposition with independent elements（IE-EMD） combining the advantages of independent component analysis algorithm and empirical mode decomposition algorithm was used to decompose the noise sources into a series of independent component by using a single observation signal to ensure the independence of isolated sound sources. At the same time, the wavelet transform technique was used to analyze the independent components and the corresponding relations between the different generator noise sources and independent components were determined by the time-frequency analysis results which was verified by experiments. Results show that these independent components correspond to valve train noise, balancing shaft drive gear noise, timing gear noise and intake air noise.

关 键 词：内燃机 发电机组 盲源分离 独立成分分析 经验模态分解 小波变换 

分 类 号：TK411.6[动力工程及工程热物理—动力机械及工程]