论文部分内容阅读
在一些应用场合,前馈有源噪声控制系统中次级源产生的声信号会反馈至参考传声器,影响参考信号质量和系统稳定,导致控制性能下降。引入了等效次级路径的概念,并通过等效次级路径与实际路径的相位偏差分析存在声反馈时的收敛性能。若某些频率的相位偏差大于90°,则这些频率附近将较难收敛,降噪性能下降,甚至导致系统不稳定。通过仿真和实验对单指向传声器声学方法、自适应滤波u型最小均方差(FuLMS)算法、反馈中和算法和在线建模算法共4种解决声反馈问题的方法的性能进行了比较。结果表明,4种方法都能提高存在声反馈时的前馈有源噪声控制系统的性能,有效解决声反馈引起的问题,但各有优缺点。单指向传声器方法最为方便,但低频指向性较差。FuLMS算法运算量较低,但不能保证收敛。反馈中和算法性能最好,但当系统时变时鲁棒性较差。在线建模算法不需要额外滤波器,但由于参数调节复杂,降噪性能稍差。
In some applications, the acoustic signals generated by the secondary sources in the feedforward active noise control system are fed back to the reference microphone, affecting the quality of the reference signal and the system stability, resulting in degraded control performance. The concept of equivalent secondary path is introduced, and the convergence performance of acoustic feedback with existence of acoustic feedback is analyzed by the phase deviation between the equivalent secondary path and the actual path. If the phase deviations of some of the frequencies are greater than 90 °, it will be more difficult to converge near these frequencies, the noise reduction performance will be degraded, and even the system will be unstable. The performance of four methods to solve the acoustic feedback problem are compared through simulations and experiments, including single-finger microphone acoustic method, adaptive filtered U-mean square error (FuLMS) algorithm, feedback neutralization algorithm and online modeling algorithm. The results show that all the four methods can improve the performance of the feedforward active noise control system in the presence of acoustic feedback and effectively solve the problems caused by acoustic feedback, but each has its own advantages and disadvantages. Single-point microphone method is the most convenient, but poor low-frequency directivity. FuLMS algorithm is computationally intensive, but can not guarantee convergence. The feedback neutralization algorithm performs best, but is less robust when the system is time varying. The online modeling algorithm does not require additional filters, but the noise reduction performance is somewhat poor due to the complex parameter tuning.