Abstract:
Aiming at the long-distance detection requirement of target orientation recognition for small-sized devices, this paper studies the problem of long-distance sound source localization using a small-aperture acoustic array mounted on small-sized devices. Based on the localization framework of Time Difference of Arrival (TDOA) estimation, a waveform threshold time delay estimation method for gunshot impulse waves is proposed. This method intercepts the characteristic points on the rising edge of the signal through a threshold and suppresses environmental noise interference by combining short-time energy detection, effectively overcoming the sensitivity of the traditional Generalized Cross-Correlation (GCC) method to waveform distortion under the conditions of low signal-to-noise ratio and small aperture. By constructing a four-element three-dimensional array with an aperture of 0.26 meters and an outdoor experimental platform 200 meters away, the localization performances of the Generalized Cross-Correlation method, the peak time delay method, and the proposed method in this paper are compared and analyzed. The experimental results show that, in the shooting range environment, the average value of the absolute error of the azimuth angle estimation of the proposed method is 1.6°, and there is no obvious deviation. The joint simulation of the algorithm further verifies that when the distance to the target is greater than 150 meters, the azimuth angle error of the system can be stably controlled within 3°. The experimental data indicate that, in an open space scenario, this sound source localization system can effectively reduce the azimuth estimation error, and its localization performance indicators are significantly better than those of traditional methods.