基于双曲正弦函数的改进变步长自适应滤波算法

Improved Variable Step-size Adaptive Filtering Algorithm Based on Hyperbolic Sine Function

针对基于双曲正弦LMS自适应滤波算法中步长函数过早达到稳态及计算开销问题,给出了一种基于双曲正弦改进变步长LMS自适应滤波算法。该算法在满足步长因子调整原则的基础上,通过步长因子与瞬时误差信号之间建立一种新的非线性函数关系,以改变步长函数底部值,避免算法过早达到稳态状态。基于该线性函数关系调整参数,消除指数运算,降低算法的计算量。仿真测试结果表明,改进算法具有较快收敛速度和较好的抗噪声能力。

In view of the problem that the step-size function in the hyperbolic sine LMS adaptive filtering algorithm suffers from steady state and computation overhead,an improved variable step-size LMS adaptive filtering algorithm based on hyperbolic sine function is presented. Based on the adjustment principle to meet the step-size factor,the algorithm establishes a new nonlinear function relationship between the step size and the instantaneous error signal,this is used to change the bottom value of the step-size function,and avoid the algorithm prematurely reaching the steady state; and based on the linear relationship,the computational overhead of the algorithm is reduced by adjusting the parameter to eliminate exponent arithmetic.The test results show that the improved algorithm has faster convergence speed and better anti-noise capability.