摘要
目前自适应光学系统中最常用的控制方法是比例-积分-微分控制器(PID),但该方法调参复杂,且在面对大气湍流等复杂动态干扰时,其固定的权重参数不足以满足高效的校正需求。本文提出基于单神经元PID控制器的波前校正算法。该算法结合了单神经网络的自适应学习能力和PID控制的优势,通过在调节权重参数,更适合动态环境下对畸变波前进行校正。以61单元的单压电变形镜为波前校正器,以不同强度的湍流像差作为校正对象,并搭建自适应光学闭环控制系统模型。仿真结果显示,当湍流像差的幅值和频率发生改变后,单神经元PID控制器收敛效果明显强于PID控制器。实验结果表明,在不同湍流强度下,单神经元PID控制器的校正性能明显优于PID控制器且离散程度也更小,说明该算法在校正畸变波前时更具有实用性。
Currently,the most prevalent control method employed in adaptive optics systems is the Proportional-Integral-Derivative controller.Nevertheless,the process of tuning its parameters is intricate,and its static weighting parameters are inadequate for fulfilling the rigorous correction requirements when confronted with intricate dynamic disturbances,notably atmospheric turbulence.Consequently,a wavefront correction algorithm grounded in a single-neuron PID controller is proposed.This algorithm integrates the adaptive learning prowess of a single neural network with the merits of PID control.By dynamically adjusting its weighting parameters,it is more suitable for compensating for distorted wavefronts in dynamic environments.A 61-element single piezo-electric deformable mirror serves as the wavefront corrector,and turbulence-induced aberrations of varying intensities constitute the objects of correction.Additionally,a closed-loop adaptive optics control system model is established.Simulation outcomes indicate that as the amplitude and frequency of turbulence-induced aberrations fluctuate,the convergence performance of the single-neuron PID controller surpasses that of the traditional PID controller markedly.Experimental results further corroborate that across different turbulence intensities,the single-neuron PID controller exhibits a significantly superior correction performance,and a smaller measure of dispersion.This suggests that this algorithm possesses greater practicality when it comes to correcting distorted wavefronts.
作者
葛鑫磊
李艳
葛晨曦
马剑强
GE Xin-lei;LI Yan;GE Chen-xi;MA Jian-qiang(Faculty of Mechanical Engineering and Mechanics,Ningbo University,Ningbo 315211,China)
出处
《激光与红外》
北大核心
2025年第8期1272-1278,共7页
Laser & Infrared
基金
国家自然科学基金项目(No.52475109)资助。
关键词
自适应光学
单神经元PID控制器
PID控制器
大气扰动校正
变形镜
adaptive optics
single neuron PID controller
PID controller
atmospheric disturbance correction
deformable mirror
