运动声源识别的动态波叠加方法研究

为更好地识别运动声源并解决声源识别中的虚假声源问题,基于运动声源短时信号的Doppler频移特性, 建立运动声场的短时波叠加关系, 利用波束形成方法对声源点进行预估, 基于预估建立起多运动声源的动态叠加方程, 进一步通过波叠加方程的求解进行声源的计算, 从而创建一种可以用于运动声源识别的动态波叠加方法. 该方法可以有效识别运动声源, 将波叠加方法扩展到了运动声源测量领域, 并在不增加传声器数量以及改变阵列形式的条件下有效抑制运动声源重建中的旁瓣效应, 解决运动声源识别中的虚假声源问题. 仿真及实际运动声源的测量试验结果证明了该方法的有效性.

Dynamic wave superposition method for moving sound sources

In order to improve the identification of moving sound sources and solve the ghost source problem, the short-time wave superposition relation for moving sound source is established according to the Doppler frequency shift in short time, the equivalent sound source is pre-estimated by beam-forming, based on the pre-estimation result, the wave superposition equation is developed and the sound source is calculated by solving the equation. Then the dynamic wave superposition method for moving sound source is obtained. This method extends the wave superposition method to the case of the moving sound sources and solves the ghost source problem in moving sound source identification without adding extra microphones or changing the layout of microphone array. This method is used in the experiment to measure the moving sound sources, and satisfactory results prove the effectiveness of the method.