Study Of Sound Source Identification Method Based On Inner Product

In this paper,the principle of inner product correlation is used for sound source identification.The signal frequency and amplitude are diagnosed through Fourier transformation.According to the microphone array order,integrate the measured sound pressure into vector form.Set up virtual point source sound pressure vector according to signal frequency and the point source radiation theory,normalize the virtual sound pressure vector to unit vector.Do product operations between measured sounds pressure vector and virtual unit vector.The position of target sound source been identified when the inner product modulus reaches maximum according to Cauchy-Schwartz inequality.The process of the sound source identification method based on inner product are deduced,simulation results show that the method can identify the sound source position and intensity accurately.In order to extend this method to practical engineering,the study of the characteristics of the method are carried out from three aspects include recognition resolution?positioning accuracy ?side-lobe suppression and so on.Under the same microphone numbers and the same array size,contrast the identification effect between cross array,circular array and rectangular array.Contrast the location identification effect in different sound source frequency,from high frequency to low frequency.Analyzed the identification effect of microphone array size,interval and measuring distance by using plane rectangular microphone array,prove that the sound source identification method based on inner product has good recognition effect in different sound source frequency,different microphone array size,different microphone interval and different measuring distance.The identification effect is discussed also when the sound pressure is disturbed by random noise,simulation results show that the method has good anti-noise ability.Under the premise of superposition principle,discuss the identification effect when there is a sound source near the target sound source which has same frequency.Simulation example shows that the mutual interference between sound sources is gradually enhanced and the weight of the two sound pressure vector is equivalent to the maximum of the inner product modulus when the distance between the two sources is gradually reduced,in this condition,two sound sources cannot be identified by the inner product iterative subtraction operation.In order to eliminate the mutual interference between two closer distance sound sources,improve recognition accuracy,a new method of sound source identification based on standard orthogonal system is proposed.In this method,according to the diagnosed source frequency,structured two virtual point source,the sound pressure values in frequency domain of the two virtual point sources are calculated respectively,according to the array order,the sound pressure values in the frequency domain of each array element are processed vector,change the two virtual sound pressure vectors into standard orthogonal systems by Schmidt orthogonalization.Do the inner product between measured sound pressure vectors andthe each vector of the standard orthogonal system,calculate the sum of inner product modules,established the objective function of the virtual sound source position coordinates,according to Bessel inequality,when the measured sound pressure vector is linearly with the standard orthogonal system,objective function research maximum.Therefore,the identification of two sound sources transformed into search maximum value of the objective function.The simulation examples proved that this method can eliminate the mutual interference between two point sources effectively,improved the accuracy of sound source identification method.The experiment of sound source localization are carried out in semi anechoic room,include single point source and two-point source localization.Using B&K 3560-D data acquisition system and rectangular microphone array for sound pressure data collection,import sound pressure data into MATLAB combine with the method in this paper for sound source identification,experimental results show that the sound source identification method based on inner product can identify the location and sound source intensity effectively,has good engineering application value.