基于线性数据重构的天线动力学模型辨识方法

Identification Method of Antenna Dynamic Models Based on Linear Data Reconstruction

射电望远镜天线伺服控制系统中的非线性特性,对系统动力学特性辨识有着显著的影响,会提高辨识难度,增加辨识模型的复杂程度.系统非线性特性的测量与补偿也会增加系统辨识工作量.针对上述问题,提出了一种基于非线性采样数据的线性重构方法,用于动力学特性建模.通过提取原采样数据的相位与幅值,对受到噪声与非线性畸变影响的系统采样数据进行线性重构,降低待辨识模型的复杂度.搭建了半实物实验平台,以平台实际采样为基础,重构线性数据,利用奇异值法与自回归神经网络评估并辨识平台动力学模型.实验结果表明,建模数据奇异值拐点从100阶下降至40阶,仅用10个神经网络节点200次训练即实现了模型辨识.

The nonlinear characteristics of radio telescope servo control system have a negative significant influence on the system dynamics characteristics identification.The measurement and compensation of system nonlinear characteristics will also increase the workload of system identification.In this research,a linear reconstruction method based on nonlinear sampling data is proposed to model dynamic characteristics.By extracting the phase and amplitude of the original sampling data,linear reconstruction of the system sampling data influenced by noise and nonlinear distortion is carried out to reduce the complexity of the model to be identified.A semi-physical experiment platform was built.Based on the actual sampling data of the platform,the linear data were reconstructed,and the dynamics model of the platform was evaluated and identified by singular value method and autoregressive neural network.The experimental results show that the singular value inflection point is reduced from 100 order to 40 order,and model identification is achieved with only 200 trainings of 10 neural network nodes.