基于改进型粒子群算法的有源噪声控制

Active noise control based on modified particle swarm optimization algorithm

为提高基于粒子群算法的有源噪声控制(ANC)系统的性能,提出一种改进型重新初始化粒子群算法(MRPSO)。该算法充分利用粒子个体最优信息,并动态改变其惯性权重,从而增强了种群的多样性,提高了算法的收敛速度和全局优化能力。针对ANC系统的时变特性,该算法通过重新初始化粒子以应对声通道的突变。以对误差信号的逐个采样为基础,介绍了基于MRPSO算法的有源噪声控制方法。该方法无须估计次级声通道,但可以有效降低噪声信号,提高信噪比。通过与已有算法的比较,结果表明MRPSO算法在全局收敛速度和优化精度上有显著的提升,同时,MRPSO算法应对声通道突变的能力也优于其他两种算法。

To enhance the performance of the active noise control （ANC） system based on particle swarm optimization algo- rithm,this paper proposed a modified reinitialized particle swarm optimization （MRPSO） algorithm. By taking full advantage of individual optimal information of all particles and dynamically changing the inertia weight, it enhanced the diversity of popula- tion and improved the convergence speed and global optimization ability. Considering the tirae-varying characteristics of ANC system,it reinitialized the particle to deal with the mutations of acoustic path. A MRPSO-based ANC method,worked on a sam- ple-by-sample principle without estimating the secondary path, could reduce the noise and improve the SNR. Simulation results show that the MRPSO algorithm can converge faster and more accurately, while it also has better ability to suit to the mutations of acoustic path than the other two algorithms.