基于进化规划的自适应IIR滤波

Adaptive IIR Filtering Based on Evolutionary Programming

针对基于梯度下降算法的自适应IIR滤波器（AIIRF）具有潜在的不稳定性和性能指标函数容易陷入局部极小而导致性能下降等问题，本文将进化规划用于AI－IRF的优化设计，不仅解决了AIIRF系统稳定性问题，而且有效地实现了滤波器性能指标函数的全局寻优和快速收敛，同时允许大动态范围的输入号。计算机仿真结果验证了基于进化规划算法的AIIRF的性能优于基于梯度算法的AIIRF，尤其对高阶、极点靠近单位圆的自适应IIR滤波器。

Adaptive IIR filtering (AIIRF) based on gradient algorithms suffers from potential instability,making the performance target function liabIe to fall down to local minimum and loading to a drop of per formance. Evolutionary programming applied to AIIRF problems can provide guaranteed filter stability.AIIRF based on evolutionary prDgramming can reach the global optimization of objective function and converge quickly. Experimental results show that AIIRF based on evolutionary programming perform is much better than AIIRF based on gradient algorithms, especially in the case where poles of filter system are close to the unit circle and for high-order filter problems.