基于独立成分分析与声阵列技术的汽油机噪声源识别

Identification of Gasoline Engine Noise Source Based on Independent Component Analysis and Acoustic Array Technology

采用基于声信号的内燃机测试分析技术对汽油机的噪声源识别展开研究,以某车用汽油机为研究对象,采用独立成分分析的方法对九点噪声信号进行盲源分离,分析各独立成分的幅值和频率随时间的变化特性,并结合汽油机的结构和噪声辐射机理,分析各独立成分的产生原因,以此来识别噪声源。同时,利用声阵列法在特定频率下对汽油机进行可视化声源定位测试。研究结果表明:利用独立成分分析的方法,可以识别汽油机的燃烧噪声、空压机驱动轮的机械啮合噪声及发电机的电磁噪声。并利用声阵列测试技术验证了盲源分离结果的可靠性。

Identification of gasoline engine noise source was studied based on acoustic signal test and analysis. With an automotive gasoline engine as a study object, the independent component analysis （ICA） was used to make the blind source separation （BSS） of noise signals at nine points, and their independent components （ICs） were obtained. The time-amplitude and time-frequency characteristics of each IC were analyzed, and the causes that make independent components take place were investigated with the combination of the gasoline engine structure and the noise radiation principle to identify each IC corresponding to the noise source. Meanwhile, the acoustic array technology was used to carry out a visualization experiment of noise source identification at a particular frequency. The result shows that the ICA can be used to identify combustion noise, mechanical noise of air compressor sprocket and electromagnetic noise of alternator; moreover, the acoustic array experiment can also be used to verify the result of blind source separation.