基于算法优化的中国高速动车组车外噪声源识别研究

为提高我国高速动车组车外噪声源识别分辨率,获得更准确的噪声源分布特征,对传统的噪声源识别波束形成算法进行多普勒效应消除算法和基于快速傅里叶变换的非负最小二乘迭代反卷积算法(FFTNNLS)的优化,并基于优化后的算法测试我国某新型动车组以不同速度通过桥梁线路区段时的车外噪声源分布。结果表明:算法优化后动车组车外噪声源识别分辨率大幅提高;动车组高速运行时,声能量主要集中于受电弓、转向架和头车排障器等区域;动车组运行速度由200km·h-1提高至350km·h-1,车辆下部区域声功率占比由91.3%降至78.9%,车体区域由6.5%升至11.5%,受电弓区域由2.2%升至9.6%。算法优化后得到的动车组车外噪声源的定位更加准确、频谱特征更加明显。

External Noise Source Identification of Chinese High-Speed EMU Based on Algorithm Optimization

In order to improve the resolution of the external noise source identification for Chinese high speed EMUs and obtain more accurate distribution characteristics of noise sources,the traditional beam forming algorithm for noise source identification was optimized by the Doppler effect elimination method and the FFT Non-Negative Least Squares(FFT-NNLS)deconvolution algorithm.Based on the optimized algorithm,the distribution of external noise sources of a new type Chinese EMU passing through bridge sections with different speeds was tested.Results showed that the resolution of external noise source identification for EMU was greatly improved after the optimization of the algorithm.The acoustic energy was mainly concentrated in pantographs,bogies and head cars when the EMU was running at high speed.As the running speed of the EMU was increased from 200 to 350 km·h^-1,the sound power ratio in the lower part of the vehicle decreased from 91.3% to 78.9%,the ratio of the carbody area increased from 6.5% to 11.5%,and the ratio of the pantograph area increased from 2.2% to 9.6%.After the algorithm was optimized,the locations of the external noise sources of EMUs were more accurate and the spectrum characteristics were more obvious.