摘要
针对大尺寸电磁振镜系统,提出“递推阻尼最小二乘—差分进化”算法进行系统辨识。首先通过递推阻尼最小二乘法对该系统进行初步模型辨识后,得到相对准确的种群范围,在此辨识基础上选择小范围的种群作为差分进化算法的初始化种群,经过反复变异、交叉和选择,最终辨识出待辨识参数的最优值。通过对该方法进行仿真辨识和实验验证,结果表明,在收敛速度基本不变的情况下,辨识结果波动更加平稳,辨识精度进一步提高。证明优化后的种群范围,可在其他参数不变的情况下,辨识精度相对于传统的差分进化算法的可提高1×10^(5)。实验结果表明,该算法的辨识结果精确,在振镜伺服控制领域有一定价值。
For large-size electromagnetic mirror systems,a"recursive damped least squares-differential evolution"algorithm is proposed for system identification.First,the system is initially identified by the recursive damped least squares method,and a relatively accurate population range is obtained.On the basis of this identification,a small population is selected as the initial population of the differential evolution algorithm.After repeated mutation,crossover and selection,Finally,the optimal value of the parameter to be identified is identified.Through simulation identification and experimental verification of this method,the results show that under the condition that the convergence speed is basically unchanged,the fluctuation of the identification result is better and stable,and the identification accuracy is further improved.It is proved that the optimized population range can improve the identification error index compared with the traditional differential evolution algorithm under the condition that other parameters remain unchanged.The results show that the algorithm is accurate,and there is a certain value in the field of gammell servo control.
作者
张暾
孟立新
张立中
白杨杨
张凯强
范涛
ZHANG Tun;MENG Lixin;ZHANG Lizhong;BAI Yangyang;ZHANG Kaiqiang;FAN Tao(College of Mechatronic Engineering,Changchun University of Science and Technology,Changchun 130022,China;Key Laboratory of Air-to-ground Laser Communication and National Defense,Changchun University of Science and Technology,Changchun 130022,China)
出处
《激光杂志》
CAS
北大核心
2022年第2期7-14,共8页
Laser Journal
基金
吉林省教育厅科学技术研究项目(No.JJKH20210836KJ)。
关键词
激光通信
模型辨识
递推阻尼最小二乘法
差分进化算法
大尺寸电磁振镜系统
laser communication
model identification
recursive damping least squares
differential evolution
large-size electromagnetic mirror system
