多声源单全息面直接声场分离与清晰化方法研究

Sound field separation, could restore the free field of target sound source in non-free acoustic field, is a method in the source identification field. In order to solve the defect of conventional sound field separation method, such as long measurement time and high test cost, and the aim to restore the free field radiated by target source simply, quickly, and accurately, this item study on direct separating multi-sources with single surface and the sharpness of the free field radiated by target source. Based on wave superposition NAH technique with only once measurement data on holography surface, the separation process of target source in multi-sources is studied. In order to make the separation result more clearly, through analyzing the conventional regularization parameter methods to obtain the universal law of each way, such as the characteristic of application condition, stability and calculation time, a method called self-adaption regularization parameter method is proposed. To get simply measurement system and low-cost, the array shape and the number of measuring points on holography surface would be optimized. Finally, experiment with simple sources will verify the separation method, which would apply the theoretical and practical basis for noise reduction processing and fault diagnosis under multi-sources condition.

声场分离是声源识别领域中非自由声场环境下还原目标声源自由声场的一种方法，可有效降低其他声源对目标声源的干扰。为解决传统声场分离方法在实际应用中的测量时间长、测试成本高等问题，以简单、快速、准确地还原目标声源的自由声场为目标，围绕多声源单全息面直接声场分离以及目标声源的自由声场清晰化问题展开研究。基于波叠加NAH技术，研究仅需一次声压测量即可实现目标声源在多声源中的分离方法；探明不同正则化参数选取方法的适用条件、稳定性、计算时间等特点的一般规律，提出一种自适应正则化参数选取方法使声场分离结果更清晰化；优化阵列形状、全息面测点数目，达到简化测量系统、降低测试成本的效果；进而进行简单源的声场分离实验研究，为实际工程中多声源情况下的降噪处理和故障诊断等问题提供理论与应用基础。

声场分离方法可有效消除相干声场中干扰源对目标声源的影响。等效源法近场声全息技术由于声场重构精度高和可对任意形状的振动体进行分析的特点，广泛应用于声场分离方法中。测量时需要一种传感器的两个测量面或者两种传感器的单测量面的声场分离方法，其测量成本较高；而仅需一种传感器的单全息面测量时，往往通过数据内插或者外推补充全息面数据，需较长的运算时间。根据等效源法近场声全息技术在理论值和测量值之间的数据传递关系，提出单全息面直接分离多个声源的方法。通过数值仿真，分析了测量点数、测量面距离、等效源数目和测量面形状在声源频率为200~3000Hz范围内对声场分离结果的影响。结果表明，对于网格阵列，测量点数在7*7到15*15范围内在整个分析频率内的声场分离结果精度近似，即采用7*7个测点即可达到足够的计算精度；而频率在500Hz以下时，由于传递矩阵列向量具有更高的相关性，声场分离误差相对增加；在测量点数低于7*7时，高的分离误差有向频率增加方向移动的趋势；测量面距离从0.05m到0.15m对于声场分离精度有较低的影响，误差相对有增加的趋势；等效源数目从22到112变化时分离结果精度基本一致；轮形阵列、圆形随机阵列、矩形随机阵列和网格阵列中，18通道轮形阵列和随机阵列在1000Hz以上时，具有和7*7网格阵列近似的分离精度，低于1000Hz时，相对误差增加。通过采用扬声器作为声源的声场分离实验，发现多声源直接声场分离方法可有效分离目标声源，为实际还原目标声源的自由声场提供了理论基础。