摘要
针对动态混合轴承信号盲分离问题,提出一种基于遗传机制改进的自适应惯性权重粒子群(GA-AW-PSO)方法。该方法以分离信号负熵作为目标函数,依据粒子适应度差值自适应调节惯性权重,以减少无效迭代次数;同时引入遗传杂交机制,增加了种群的多样性,有利于处理动态混合信号;此外,为降低算法复杂度将正交矩阵表示为参数化的形式。仿真表明,该方法对动态混合的模拟机械信号盲分离时性能优于传统算法;能分离实际动态轴承信号,达到了故障检测目的。
The adaptive particle swarm optimization based on genetic mechanism（GA-AW-PSO） is proposed, aiming at blind source separation for dynamic hybrid bearing signals. The negentropy of separated signal is regarded as an objective function. The inertia weight is adjusted adaptively to reduce the invalid iterations according to the fitness difference. The introduction of genetic mechanism can increase diversity and is helpful for dynamic signal processing. The parameterized representation of orthogonal matrices can reduce the complexity of the algorithm. The simulation results show that the proposed method is superior to traditional blind source separation for the dynamic mechanical hybrid analog signal. It can effectively separate the actual dynamic bearing signal and reach the purposes of fault detection.
作者
张天骐
马宝泽
强幸子
全盛荣
Zhang Tianqi;Ma Baoze;Qiang Xingzi;Quan Shengrong(Chongqing Key Laboratory of Signal and Information Processing （CQKLS ＆ IP）, Chongqing University of Posts and Telecommunications （CQUPT）, Chongqing 400065, Chin)
出处
《系统仿真学报》
CAS
CSCD
北大核心
2018年第6期2306-2314,共9页
Journal of System Simulation
基金
国家自然科学基金(61671095
61371164
61275099)
关键词
盲源分离
粒子群
遗传杂交
轴承故障信号
blind source separation
particle swarm optimization
genetic hybrids
bearing fault signal
