基于LS-AGA伺服系统连续模型参数辨识

Identification of Continuous-time Model Based on LS-AGA Algorithm for Servo System

为了解决伺服系统连续模型参数辨识不足的问题,提出一种间接的连续模型辨识算法,建立了伺服系统的动力学模型。首先使用最小二乘法(LS算法)来辨识出系统的离散模型参数,其次通过自适应遗传算法(AGA算法)对离散参数进行Z-S变换,获得连续系统模型参数。根据LS-AGA辨识算法设计了辨识实验,采用扫频正弦模拟量电压作为输入信号对系统中的连续参数模型进行辨识,并对所获得的连续模型进行了分析和验证。研究结果表明,辨识模型的预测输出与实际实验输出匹配度较高,该连续模型能够反映伺服系统的特性。

In order to overcome the disadvantages of continuous-time servo system identification,an indirect identification of continuous-time models via discrete-time forms was proposed.In this indirect approach,a discrete-time model of the servo system is first estimated by least squares approaches and a continuous-time model is then deduced from it by Z-S transformation via adaptive genetic algorithm.After designing the experiment,by using a swept-frequency sinusoid waveform of the analog input voltage according to the LS-AGA identification algorithm,the unknown parameters of the continuous-time servo model are identified.Finally,the established servo model is analyzed and validated.The results show that the prediction output and the actual experiment output match well,and the continuous-time model can reflect the characteristics of the servo system.