针对等效源法近场声全息声场重建中Tikhonov最优正则化参数选取问题,本文引入Bayesian正则化准则法进行求解,并与GCV法和L-curve法进行仿真对比分析。结果显示,Bayesian正则化准则法在宽频带内有优异的重建性能。进一步的计算及试验表明,该方法能够稳定准确地实现声场重建。 Aiming at the problem of Tikhonov optimal regularization parameter selection in the reconstruction of near-field acoustic holographic sound field using equivalent source method, the Bayesian regularization criterion method is introduced to solve the problem. The GCV and L-curve methods are compared and analyzed. The results show that the Bayesian regularization criterion method has excellent reconstruction performance in a wide frequency band. Further calculations and experiments show that this method can stably and accurately reconstruct sound field.