基于神经网络逆模型在线迭代优化的三轴标准振动台解耦控制
DOI:
CSTR:
作者:
作者单位:

太原理工大学机械工程学院太原030024

作者简介:

通讯作者:

中图分类号:

TH71

基金项目:

国家自然科学基金项目(52375544)、山西省研究生实践创新项目(2025SJ125)资助


Decoupling control of tri-axial standard vibrator based on online iteration optimization of neural network inverse model
Author:
Affiliation:

College of Mechanical Engineering, Taiyuan University of Technology, Taiyuan 030024, China

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    为确保三轴振动传感器的校准精度,需由三轴标准振动台输出低耦合、低失真的振动信号。然而,由于三轴标准振动台结构的复杂性和非线性特性,其输出信号存在显著的残余耦合和谐波失真。为此,提出一种神经网络逆模型在线迭代解耦控制方法,以实现轴间解耦并提高输出信号精度。首先,以簧片式三轴标准振动台为研究对象,对其解耦结构及轴间运动耦合特性进行了理论分析,并量化了正交振动抑制比和谐波失真度。然后,利用神经网络构建了三轴标准振动台系统的逆模型,并采用前馈串联方式对原系统实施控制,初步改善了轴间耦合干扰和谐波失真。在此基础上,为更精准表征三轴标准振动台的动态耦合特性,引入在线迭代学习机制,通过动态更新样本对逆模型进行周期性迭代优化,逐步提升其拟合精度,从而实现了对三轴标准振动台轴间耦合和谐波失真的精确控制。最后,基于AD7606B搭建了振动采集分析系统,并集成构建了三轴标准振动台实验测试系统。实验分析结果表明,神经网络逆模型在线迭代解耦控制方法可将三轴标准振动台的正交振动抑制比控制在2%以内,同时将谐波失真度控制在1%以内,验证了该解耦控制方法的有效性,为多轴振动测试系统的高精度解耦控制提供了技术参考。

    Abstract:

    To ensure the calibration accuracy of tri-axial vibration sensors, it is necessary for the tri-axial standard vibrator to output low-coupling and low-distortion vibration signals. However, due to the complexity and nonlinearity of the tri-axial standard vibrator structure, the output signals exhibit significant residual coupling and harmonic distortion. To address this issue, this paper proposes a neural network inverse model online iterative decoupling control method to achieve decoupling between axes and improve the accuracy of the output signals. Firstly, taking the leaf-spring-type tri-axial standard vibrator as the research object, the decoupling structure and the coupling characteristics of inter-axis motion were theoretically analyzed, and the orthogonal vibration suppression ratio and harmonic distortion were quantified. Then, the inverse model of the tri-axial standard vibrator system was constructed using the neural network, and the original system was controlled by feedforward series method, which initially improved the inter-axis coupling interference and harmonic distortion. On this basis, to more accurately characterize the dynamic coupling characteristics of the tri-axial standard vibrator, an online iterative learning mechanism was introduced. By dynamically updating the samples, the inverse model was periodically iteratively optimized to gradually improve its fitting accuracy, thereby achieving precise control of the inter-axis coupling and harmonic distortion of the tri-axial standard vibrator. Finally, a vibration acquisition and analysis system was built based on AD7606B, and an experimental test system for the tri-axial standard vibrator was integrated and constructed. The experimental analysis results show that the neural network inverse model online iterative decoupling control method can control the orthogonal vibration suppression ratio of the tri-axial standard vibrator within 2% and the harmonic distortion within 1%, verifying the effectiveness of the decoupling control method and providing a technical reference for high-precision decoupling control of multi-axis vibration test system.

    参考文献
    相似文献
    引证文献
引用本文

张旭飞,武艺凡,王硕,魏鑫,许佗.基于神经网络逆模型在线迭代优化的三轴标准振动台解耦控制[J].仪器仪表学报,2026,47(4):213-223

复制
分享
相关视频

文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2026-06-08
  • 出版日期:
文章二维码