Sound Source Identification Method By Using A Single Moving Microphone And Inner Product Calculation

Noise control is based on accurately noise source identification.Only by identifying the location of sound source and studying its distribution can noise be better controlled.Microphone array technology generally has the problem of phase mismatch error and frequency response inconsistency.Furthermore,a great deal of calibration work should be finished before measurement in order to achieve better identification accuracy.In this paper,a single moving microphone combined with inner product correlation is used to locate the sound source,based on this,intensity and initial phase of sound source information are calculated.The technical details and application characteristics of the method are discussed,identification performance of the method is verified by simulation calculations and experiment.The sound source identification method in this paper is based on inner product calculation.The principle of inner product method is as follows: Time domain sound pressures of each array element is transformed into frequency domain and integrated into sound pressure vector in sequence.A virtual sound source with the same frequency is constructed in the region,and the virtual sound pressure vector is obtained by the same processing method,then inner products of two vectors are calculated.Sound source is located by searching for the maximum of the inner products.Based on the inner product algorithm,a single moving microphone is used to identify the sound source.Firstly,the static microphone is used to collect sound pressure signals and then identify the frequency of the sound source.Microphone is arranged to move along the given trajectory at low speed while collecting the sound pressure signals.The time-domain sound pressures are segmented and converted into a frequency-domain value by time-frequency conversion,then discrete time-domain value of each segment is integrated into a vector in sequence.Virtual point sources are constructed,and the sound pressure of the virtual source is collected and processed by the analog microphone in the same way.The sound pressure vector corresponding to each virtual point source is obtained and normalized.The source can be located by searching the maximum inner product modulus of two vectors,and then the intensity and initial phase of the source can be calculated.This method is applied to the identification of point source in plane and space,meanwhile,the anti-noise ability is investigated.The simulation results show that the proposed method can effectively identify the sound source and has good anti-noise potential.The identification effects for different frequency sources at different segment time intervals are investigated,meanwhile the characteristics of the method are analyzed.After analyzing the simulation results,it is found that sound source identification results are related to segment time intervals.The identification result is biased when segment time interval is less than 6 signal cycles,and the error is mainly caused by the integration of negative frequency term.After analyzing the trend of two integral values of time-frequency transform,a half-period segmentation method is proposed to reduce the interference of negative frequency term and improve the accuracy of sound source identification.The simulation example shows that the identification results are not limited by the length of time intervals when half-period segmentation is used,and the identification results are independent of the frequency of the source and the numbers of sound pressure segments.After discussing the identification results when movement of the microphone is deviated,it is found that the smaller trajectory deviation and motion velocity deviation have no obvious influence on the identification results.Sound source identification experiment was carried out in the semi-anechoic laboratory.The non-directional spherical source was selected as the identification object,and the high-sensitivity microphone was used to collect sound pressure signals while moving along the given orbit.Sound pressure data were imported into the pre-programmed sound source identification program to observe the distribution of inner product modules and its projection map.It is found that there is a unique peak near the sound source,which indicates that a single moving microphone can effectively locate the sound source.The localization experiments for different frequency sound sources show that the method has universal applicability for different frequency bands and has good engineering practicability.