改进遗传算法径向基函数的FIR数字滤波器研究

Study of FIR Filtering Algorithm Based on the Improved Genetic Algorithm Radial Basis Function(RBF) Network

本文针对径向基函数(RBF)网络在实际应用中几种学习算法存在的缺陷和局限性,提出了利用改进遗传算法的径向基函数网络来设计FIR数字滤波器的方法,提高了全局寻优效率.为了证明这种新型混合遗传算法的有效性,本文用该算法设计了低通、高通FIR数字滤波器.采用Matlab进行仿真并给出了相应的仿真结果以验证其有效性.仿真结果表明:同其他现有的优化设计方法相比,本文研究的这种新算法可以很好地逼近理想滤波器,在通阻带内有更小的纹波,达到比较好的优化设计效果.

This paper is based on the defects and limitations of several learning algorithms in the practical application of radial basis function（RBF） network,proposing the FIR digital filters designed by RBF network genetic optimization algorithm to improve the global optimization efficiency,the optimization algorithm adopted the improved genetic algorithm as the RBFNN learning algorithm.In order to prove the effectiveness of the new hybrid genetic algorithm,the low-pass and the high-pass FIR digital filters are designed through this algorithm in this paper,using Matlab simulation and giving the corresponding simulation results to verify the effectiveness of it.The simulation results show that compared with other existing design optimization methods,the new algorithm of this paper can further approximate the ideal filters,having the smaller ripple in the passband and the stopband and achieving better optimization design effect.