Study On Moving Noise Source Identification

Noise measurement is one of the key measure methods in noise experiments. The occurrence of noise source identification technology adapts to the current noise control engineering. It provides an effective method in quality inspection and performance improvement of the Noise Vibration system. Researching on noise source identification technology, it can't be only stayed in static noise source test, but also needs to study on the characteristics of moving noise source. Especially, the existed noise accompanied by movement, has brought some difficulties in noise measurement and analysis, such as tire noise and aerodynamic noise. Most scholars at home and abroad analyzed the moving noise source by Near-field Acoustic Holography (NAH) or Beamforming. But the identification effect is not so accurate because of the limitations in the arithmetic character and practical measurement. This dissertation focuses on the analysis of the sound field radiation, around this topic to search for an effective method of the moving noise source identification and analyze sound field characteristics accurately.NAH and Beamforming are analyzed and discussed in detail as the main topic for the moving noise source identification. The corresponding solution and improvement measures are proposed against existing problems in the two methods for the moving noise source identification. On this basis, a combined measurement technique is proposed. It has the priority on both of NAH and Beamforming. Using NAH at low frequencies and Beamforming at high frequencies, the sound field distribution is gained at the entire frequency domain. The combined measurement technique is proved to be feasible and superior through theoretical analysis, mathematical modeling and experimental validation. The main study of this dissertation is completed as follows:Firstly, the problem description of acoustic radiation is established and the numerical solution is deduced. The acoustic radiation characteristics of the moving sound source are investigated in respect of the kinematic and dynamic, which provides a foundation for rebuilding acoustic radiation.The amplitude and frequency changes of Doppler signal created by the moving sound source are analyzed systematically. The method based on wavelet analysis is raised to eliminate the Doppler Effect and the feasibility is demonstrated through specific examples. It provides a new solution to eliminate the Doppler Effect.Secondly, the key problem was solved using NAH to analyze the moving noise identification technology.The mathematical description of the moving noise sources identification based on NAH is presented. Aiming at the microphone layout problem, the optimization layout of microphones based on the genetic algorithm is proposed. It is proved that the array generated by optimized method conforms to NAH. In addition, it doesn't affect resolving power and array performance, which can reduce the measurement cost. The exemplification illustrates that NAH can identify the moving noise source comparatively with the superior low-frequency resolution, but the result is not so satisfactory at high frequencies.Furthermore, the moving noise source identification technology based on Beamforming is reviewed and the relevant algorithm is improved. The delay, the weight and summation formula of Beamforming is deduced. According to maximum likelihood criteria, the maximum likelihood estimation of the useful signal is achieved to select reasonable weighted coefficient of Beamforming. The theoretical description and numerical realization show that Beamforming can identify the moving noise source comparatively with the superior high-frequency resolution, but identification effect is not so satisfactory at low frequencies.According to the above analysis, the moving noise source identification method based on the combined measurement technique is approved. Comparing to the spatial resolution between NAH and Beamforming, it is easy to see that: NAH has the superior in low-frequency resolution. Unfortunately, it is difficult to meet practical application in heigh-frequency using NAH. The good news is that Beamforming has the superior in high-frequency resolution, but it is a big challenge of using so many microphones at low frequencies to reach a certain resolution. The analysis and discussion about the array design indicates that: NAH demands a regular and grid measurement array. Beamforming does not restrict the regular degree and aperture size of the array, but the use of rectangular array can improve the maximum measurable frequency and the anti-noise ability. In order to lessen microphones and reduce the cost, a T-shape array is used to receive signals in the combined measurement technique, and it is expanded into the rectangular array signal through the correlation function theory. The examples explain that: When the sound field reconstruction accuracy is not required very high, the expansion signal of the measurement array can locate the sound source correctly. The array selection is not affected by the volume of the sound source and measurement distance. It doesn't need to reset array under different circumstances. In order to avoid the incomplete, inaccurate, and others uncertain acoustic analysing results at high frequencies and low frequencies caused by the division of frequency, measurement distance as well as reconstruction algorithm itself in the combined measurement, identification results at high frequencies and low frequencies are regarded as a number of evidences. According to the evidence theory, these evidences are synthesized and a final recognition result is obtained. Exemplification shows: The sound source identification rate has been increased by 10.3% after the integration.Finally, the accuracy and superiority of the combined measurement technique are validated in terms of experiments. For the single-speaker experiment in the semi-anechoic chamber: NAH gains a good effect at low frequencies. The identification based on Beamforming is less effective. At high frequencies, two opposite results are received. Meanwhile, the combined measurement technique has gained good results in the whole frequency range. The identification error is less than 7.6%.In the dual-speaker experiment, it is impossible to distinguish between two sound sources well based on NAH or Beamforming in excess of a certain frequencies range. But the combined measurement technique still maintains good results. The identification error is less than 8.7%. The experiments emphasize that the combined measurement technique locate the steady sound source well. In the main noise sources of the moving vehicles experiment, the combined measurement technique has good effect in the whole frequency, and the conclusions match the actual location of noise sources. The error is in the range of±0 .18m, which is better than the identification results of the literatures.The experiment shows that the combined measurement technique can identify the moving noise source correctly. Moreover, the number of microphones is much less than that of the optimized array in B&K Company. Thus, the experimental cost has been reduced and the efficiency has been improved, which provide the possibility of practical application in the project.The innovation of the dissertation consists in the improvement of NAH and Beamforming, extracting the moving noise source identification method based on combined measurement technique.The significance is that the new technique has an advantage over the past noise source identification methods not only in a wider frequencies range, but also in lower measurement cost, which has practical value in the moving noise source identification domain.