论文部分内容阅读
为消除变形镜的建模误差,提出了基于在线最小二乘支持向量机的变形镜建模方法。首先,分析了温度和驱动器非线性等因素对变形镜响应矩阵的影响;然后,介绍了最小二乘支持向量机及在线更新的原理,并将其引入97单元变形镜的集成仿真模型。根据变形镜不断更新的运行数据,最小二乘支持向量机进行在线训练和模型更新,构建当前状态变形镜的等效模型,并输出下一时刻的电压预报值。仿真结果表明:基于在线最小二乘支持向量机的变形镜建模方法摆脱了固定模型的约束,具有自适应更新的特点,稳健性好,控制电压预测精度高,有利于提高自适应光学系统的控制精度。
In order to eliminate the modeling error of the deformable mirror, a deformable mirror modeling method based on online least square support vector machines was proposed. Firstly, the influence of temperature and driver nonlinearity on the response matrix of deformable mirrors is analyzed. Then, the principle of LS-SVM and online updating is introduced and introduced into the integrated simulation model of 97-unit deformable mirror. According to the continuously updated operating data of the anamorphic mirror, the least squares support vector machine performs online training and model updating, constructs the equivalent model of the current state anamorphic mirror, and outputs the voltage prediction value of the next moment. The simulation results show that the deformation mirror modeling method based on online least square support vector machine can get rid of the constraints of the fixed model and has the characteristics of adaptive updating, good robustness and high prediction accuracy of control voltage, which is beneficial to improve the performance of adaptive optics system control precision.