基于遗传算法的信号识别技术  

Signal identification technology based on genetic algorithm

作　　者：陶继忠[1] 殷国富[1] 汪法根[2] 

机构地区：[1]四川大学制造科学与工程学院,四川成都610065 [2]中国工程物理研究院机械制造工艺研究所,四川绵阳621900

出　　处：《机械设计》2004年第8期10-11,51,共3页Journal of Machine Design

基　　金：国家自然科学基金资助项目 (50 2 751 0 0 ) ;中国工程物理研究院科学技术基金资助项目 (2 0 0 30 32 6)

摘　　要：将遗传算法应用于识别气浮式动平衡测量装置准静态信号的参数 ,提出了二次磨光法的数据预处理方法。首先采用二次磨光法消除坏点并进行快速傅立叶变换以确定信号的频率成分 ,并设定各频率成分信号的幅值和初相位的取值范围 ,然后以实测信号与识别结果的差值定义适应度 ,最后经过基本遗传操作和迭代得出最优解。仿真结果和试验实例表明 ,提出的方法精度很高 ,幅值识别误差不超过 1% ,初相位误差小于 2° ,满足了实际工程需求。Let the genetic algorithm be applied to identify the parameters of quasi-static signal of air suspension dynamic balance measuring devise, a data pre-treatment method for quadric-polishing function was advanced. Firstly, using quadric-polishing function to eliminate bad points and carrying out Fourier transformation so as to determine the frequency components of signal, and set up signal amplitude of each frequency component and value adoption range of preliminary phase, and then taking the difference between actually measured signal and result of identification to define the fitness, finally, the optimal solution is obtained through basic genetic operation and iteration. The result of simulation and living example test showed that the accuracy of proposed method is quite high with the amplitude identifying error less than 1%, and the preliminary phase error less then 2°,which satisfied the needs of practical engineering.

关 键 词：信号识别 遗传算法 最优解 动平衡 信号处理 

分 类 号：TH113.25[机械工程—机械设计及理论]