基于自适应混合结构的快速收敛函数链接人工神经网络算法研究  

Fast convergence FLANN algorithm based on an adaptive mixing structure

作　　者：李欢欢[1] LI Huanhuan(Library,Mudanjiang Normal University,Mudanjiang 157011,China)

机构地区：[1]牡丹江师范学院图书馆,黑龙江牡丹江157011

出　　处：《振动与冲击》2021年第10期180-186,共7页Journal of Vibration and Shock

基　　金：牡丹江师范学院青年培育项目(1354MSYQN014);牡丹江师范学院人文社会科学技术研究项目(YB2019015)。

摘　　要：在非线性主动噪声控制方法中,函数链接人工神经网络(FLANN)算法是最常用的算法之一。FLANN降噪量大,但是其收敛速度较慢。为解决该问题,通过一个自适应混合参数对BFXLMS算法和FLANN算法进行有效结合,提出了CBFLANN算法。在不降低FLANN降噪量的情况下,提高了其收敛速度,解决了FLANN算法无法同时实现快速收敛和低稳态误差的问题。多个仿真实验对提出的CBFLANN算法的降噪性能进行了验证,结果表明,CBFLANN同时拥有BFXLMS的收敛速度和FLANN的降噪量。该算法的提出可以为传统主动噪声控制算法难以同时兼顾收敛速度与稳态误差的问题提供解决方案,具有很强的实际应用价值。Among the nonlinear active noise control methods,the functional link artificial neural network(FLANN)algorithm is one of the most commonly used algorithms.The FLANN has a better noise reduction result,but its convergence speed is slow.In order to solve this problem,a new combined bilinear FLANN(CBFLANN)algorithm by combining the bilinear filtered-x LMS(BFXLMS)algorithm with the FLANN algorithm using an adaptive mixing parameter was proposed.Without decreasing the noise reduction capability of FLANN,the convergence speed of FLANN was improved,such that the fast convergence and low steady-state error for the FLANN were achieved at the same time.To verify the performance of the proposed CBFLANN algorithm,several simulation experiments were conducted.The results show that the CBFLANN has both the convergence rate of BFXLMS and the steady-state error of FLANN at the same time.The proposed algorithm can provide a solution to the difficulty in considering both the convergence rate and the steady-state error for the conventional active noise control algorithms.It has a strong practical application value.

关 键 词：非线性主动噪声控制 函数链接人工神经网络(FLANN) 收敛速度 稳态误差 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]