改进的分段盲源分离方法及其应用  

Improved Segmental Blind Source Separation Algorithm and Its Application

作　　者：李世龙[1] 陈建宏[1] LI Shilong;CHEN Jianhong(Taiyuan Satellite Launch Center,Taiyuan 030027,China)

机构地区：[1]太原卫星发射中心,山西太原030027

出　　处：《自动化仪表》2018年第10期79-83,共5页Process Automation Instrumentation

摘　　要：针对传统盲源分离算法大多存在收敛速度慢、分离精度低的缺点,提出了一种改进的分段盲源分离算法。将整个信号分离过程分为快速分离和精细分离两个阶段。在快速分离阶段,采用基于粒子群优化(PSO)的分离算法,通过较少的迭代次数实现信号较好的初步分离;在精细分离阶段,通过选择适当的学习速率,进一步提高信号的分离精度。通过数值仿真及试验分析,将改进的分段盲源分离算法与现有的分离性能较好的基于人工蜂群(ABC)的分离算法进行了对比分析。结果表明,改进的分段盲源分离算法具有更优异的分离速度、分离精度和稳定性。In view of the disadvantages of slow convergence speed and low accuracy in traditional blind source separation algorithm, an improved segmental blind source separation algorithm is presented. The signal separation process can be divided into two stages：the rapid separation and the precise separation. The particle swarm optimization （PSO） algorithm is applied in the rapid separation stage to get better result with less iteration times, and the signal separation accuracy is further improved by proper learning rate in the precise separation stage. By numerical simulation and experimental analysis, the improved segmental blind source separation algorithm is compared with the blind source separation algorithm based on ABC method, and the results show that the proposed algorithm is superior in the separation speed, accuracy and stability.

关 键 词：盲源分离 粒子群优化 快速分离 精细分离 故障轴承 频谱分析 

分 类 号：TH113.1[机械工程—机械设计及理论] TP181[自动化与计算机技术—控制理论与控制工程]