For further solving the contradiction of between the convergence speed and the steady-state error, and for improving the performance of adaptive filtering algorithm, the variable step size algorithm is introduced into the convex-combination adaptive filtering algorithm (CLMS), and thus a new variable step-size convex-combination least-mean-square (VSCLMS) algorithm is proposed. The algorithm, with a variable step size algorithm based on block step size, replaces the fast convergence algorithm in the original CLMS algorithm. Theoretical derivation and simulation analysis indicate that the proposed algorithm could improve the flexibility and practicability of the CLMS algorithm, and exhibits much better filtering effect than other variable step size algorithms, CLMS algorithm and proposed VSCLMS algorithm in the three aspects of convergence speed, steady state error and tracking performance.%为了进一步解决收敛速度与稳态误差之间的矛盾,提高自适应滤波算法性能,将变步长算法引入到凸组合自适应滤波算法(CLMS)中,提出了一种新的变步长凸组合最小均方算法(NVSCLMS)。该算法用一种基于块步长思想的变步长算法代替原CLMS算法中的快速收敛算法。通过理论推导和仿真对比分析:提出的算法提高了CLMS算法的灵活性和实用性,且在收敛速度、稳态误差和跟踪性三个方面展现出优于变步长算法、CLMS算法和已提出变步长凸组合算法的滤波效果。
展开▼