一種優(yōu)化動(dòng)量因子的盲分離算法論文(參考版)
2025-06-22 05:49本頁(yè)面
【正文】 References1 X. Zhu, X. Zhang, Adaptive RLS algorithm for blind source separation using a natural gradient, IEEE Signal Process. Lett. 9(12)(2002)432435.2 Coviello C M, Yoon P A, Sibul L H. Source separation and tracking for timevarying systems. IEEE Transactions on Aerospace and Electronic Systems, 2008, 44(3): 119812143 Douglas S C, Cichocki A. Adaptive step size techniques for decorrelation and blind source separation. In: Proceedings of the 32th Asilomar Conference on Signals, Systems, and Computers. Paci175。圖3 算法收斂性能(平穩(wěn)環(huán)境)Fig. 3 Average performance index versus iteration number(stationary environment) 圖4 分離矩陣偏離正交性(平穩(wěn)環(huán)境)Fig. 4 Average deviations of the separation matrix away from orthogonality(stationary environment) 圖5 相對(duì)于計(jì)算量的算法特性Fig. 5 Average performance index versus Operations (stationary environment)圖6 算法收斂性能(非平穩(wěn)環(huán)境)Fig. 6 Average performance index versus iteration number(nonstationary environment)圖7 分離矩陣偏離正交性(平穩(wěn)環(huán)境)Fig. 7 Average deviations of the separation matrix away from orthogonality(nonstationary environment)圖8 相對(duì)于計(jì)算量的算法特性Fig. 8 Average performance index versus Operations (nonstationary environment)5 結(jié)論 在LMS算法基礎(chǔ)上引入動(dòng)量項(xiàng)，然后利用投影近似和相鄰時(shí)刻代價(jià)函數(shù)相減法，得到關(guān)于動(dòng)量因子的二次函數(shù)，求極值得到最優(yōu)動(dòng)量因子，從而使代價(jià)函數(shù)最速下降，提高了算法收斂速度。(3) 平穩(wěn)環(huán)境和非平穩(wěn)環(huán)境下，在計(jì)算量相同時(shí)，與其它算法相比
公司管理相關(guān)推薦
鄂ICP備17016276號(hào)-1