多小波模糊神经网络盲均衡算法

Multi-wavelet Fuzzy Neural Network Blind Equalization Algorithm in Underwater Acoustic Channel

为了克服传统恒模算法(CMA)采用固定步长造成的收敛速度与收敛精度之间的矛盾,提出了一种基于平衡正交多小波变换的模糊神经网络盲均衡算法(MWT-FNN-BEA).该算法一方面利用模糊神经网络控制器自动调节算法的迭代步长,较好地解决了收敛速度与收敛精度之间的矛盾;另一方面利用平衡正交多小波变换对均衡器输入信号进行去相关性处理,进一步提高了算法的性能。理论分析和水声信道仿真结果表明,所提出的算法具有较快的收敛速度和较小的稳态误差,抗干扰性能好。

In order to overcome the contradiction between the convergence rate and accuracy for using fixed step in traditional constant modulus algorithm（CMA）,a fuzzy neural network blind equalization algorithm based on balance orthogonal multi-wavelet transform（MWT-FNN-BEA）was proposed.In this algorithm,the contradiction between the convergence rate and accuracy was solved by using fuzzy neural network controller to adjust the step of algorithm automatically.Furthermore,by utilizing the balance orthogonal multi-wavelet transform to the input signals of equalizer,the performance of algorithm was further improved.Theoretical analysis and simulation results demonstrate that the proposed algorithm has higher convergence rate,small steady error and good noise immunity.