中国机械工程 ›› 2025, Vol. 36 ›› Issue (11): 2574-2582.DOI: 10.3969/j.issn.1004-132X.2025.11.012

• 机械基础工程 • 上一篇    

基于超声谐振挤压膜效应的触觉纹理动态感知及实验研究

陈智博1, 李国平1(), 项四通1, 魏燕定2   

  1. 1.宁波大学机械工程与力学学院, 宁波, 315211
    2.浙江大学机械工程学院, 杭州, 310058
  • 收稿日期:2024-10-28 出版日期:2025-11-25 发布日期:2025-12-09
  • 通讯作者: 李国平
  • 作者简介:陈智博,男,1999年生,硕士研究生。研究方向为压电精密驱动
    李国平*(通信作者),男,1967年生,教授。研究方向为精密运动控制。E-mail: liguoping@nbu.edu.cn
    第一联系人:魏琼,女,1980年生,副教授。研究方向为流体传动控制、机电伺服系统设计。E-mail:20140058@hbut,edu.cn。李奕*(通信作者),男,1976年生,副教授。研究方向为流体传动控制、机电一体化。E-mail:570925693@qq.com
  • 基金资助:
    国家自然科学基金(51975517);国家自然科学基金(52075273);国家自然科学基金(51805276);浙江省自然科学基金(LGF21E050002)

Dynamic Perception and Experimental Study of Tactile Texture Based on Ultrasonic Resonance Squeeze Film Effect

Zhibo CHEN1, Guoping LI1(), Sitong XIANG1, Yanding WEI2   

  1. 1.School of Mechanical Engineering and Mechanics,Ningbo University,Ningbo,Zhejiang,315211
    2.College of Mechanical Engineering,Zhejiang University,Hangzhou,310058
  • Received:2024-10-28 Online:2025-11-25 Published:2025-12-09
  • Contact: Guoping LI

摘要:

利用调控接触界面的滑动摩擦可以感知变化的触觉纹理特征的机制,基于挤压膜效应提出了一种超声谐振触觉动态感知装置。在挤压膜的高挤压状态下,构建挤压膜效应减摩机制理论模型,通过摩擦因数标定实验建立摩擦调控性能与实际材质纹理的映射关系。研究结果表明,该装置在频率为36.314 kHz的谐振信号激励下,电压幅值在0~200 V时可实现表面摩擦因数0.295~0.807的动态调控。利用长短期记忆(LSTM)神经网络构建时序力信号与触觉纹理预测模型,客观评估装置的触觉纹理再现性能,得到该模型预测结果的平均误差为3.33%,验证了装置具有较好的触觉纹理再现效果。

关键词: 超声振动, 挤压膜效应, 摩擦调控, 触觉再现, 长短期记忆(LSTM)神经网络

Abstract:

Using the mechanism that modulating sliding friction at the contact interfaces might perceive changing tactile texture features, an ultrasonic resonance tactile dynamic perception device was proposed based on the squeeze film effect. The theoretical model of the friction-reducing mechanism of the squeeze film effect was constructed under the high squeeze factor of the squeeze film, and the mapping relationship between the friction-regulating performance and the actual material texture was established through the friction factor calibration experiments. The results show that the device may achieve the dynamic regulation of surface friction factor is as 0.295~0.807 under the excitation of 36.314 kHz resonance signals and the voltage amplitude is as 0~200 V. A LSTM neural network was used to construct a temporal force signal and tactile texture prediction model to objectively evaluate the tactile texture reproduction performance of the device, and the average error of the model prediction results is as 3.33%, which verifies the device has a good reproduction effectiveness of tactile texture.

Key words: ultrasonic vibration, squeeze film effect, friction modulation, haptic reproduction, long short-term memory(LSTM) neural network

中图分类号: