论文部分内容阅读
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。
In view of the potential instability of the adaptive IIR filter (AIIRF) based on the gradient descent algorithm and the easy entry of the performance indicator function into the local minimum resulting in performance degradation, the evolution planning is used in the optimization design of the AI-IRF not only Which solves the problem of stability of AIIRF system and effectively achieves the global optimization and fast convergence of the filter performance index function while allowing the input numbers of large dynamic range. Computer simulation results show that AIIRF based on evolutionary programming algorithm outperforms AIIRF based on gradient algorithm, especially for high-order, pole-close adaptive unit IIR filters.