基于NFxPEM算法的调制气流声源的非线性补偿  

Non-linearity compensation of air-modulated speaker based on the nonlinear filtered-x PEM algorithm

作　　者：周泽民[1] 曾新吾[1] 龚昌超[1] 田章福[1] 孙海洋[1] 

机构地区：[1]国防科技大学光电科学与工程学院,湖南长沙410073

出　　处：《国防科技大学学报》2013年第6期126-131,共6页Journal of National University of Defense Technology

基　　金：国家自然科学基金资助项目(11274387)

摘　　要：针对调制气流声源存在较强的谐波畸变,将声源系统等效为Hammerstein非线性模型,利用该模型下的预失真技术对声源进行非线性补偿研究。根据辨识的Hammerstein模型中静态非线性部分带有直流分量的特点,给出了考虑直流分量补偿的预失真算法,并用数值仿真验证了算法的准确性和直流分量补偿的必要性。在非线性补偿实验中,根据单频信号辨识得到Hammerstein模型参数,采用NFxPEM算法求得对应的预失真Wiener模型参数和预失真波形。实验结果表明,与直接发射相比,补偿发射后声波的功率谱中谐波能量有所下降,而基频能量有小幅度的上升,说明了研究思路的正确性。Aimed at the harmonic distortion problem in the air-modulated speaker（ AMS）, the AMS behavioral model was represented by a Hammerstein structure, and the research on predistortion of AMS based on this model was made. As the DC offset exists in the nonlinearity of the Hammerstein model, a predistortion algorithm considering the DC offset compensation was developed. The validity of the algorithm and ttle necessity of the DC offset compensation were verified by computer simulation. In the experiment, a single sinusoidal excitation signal was first used to identify the Hammerstein model. Then, using the identified system parameters, the NFxPEM algorithm was performed to obtain the parameters of Wiener predistorter and to predistort the excitation signal. From the experiment results, it is found that our approach is effective in reducing the harmonic power with a relatively small upgrade in the fundamental frequency power.

关 键 词：调制气流声源 HAMMERSTEIN模型 NFxPEM算法 非线性补偿 

分 类 号：O422.7[理学—声学]