Self-sensing of dielectric elastomer actuator enhanced by artificial neural network
Self-sensing of dielectric elastomer actuator enhanced by artificial neural network
复制标题
DOI：
10.1088/1361-665x/aa7e66
复制
发表时间：
2017-08
影响因子：
4.1
通讯作者：
Zhihang Ye;Zheng Chen
中科院分区：
文献类型：
--
作者：
Zhihang Ye;Zheng Chen
介电弹性体（DE）是一种柔性驱动材料，其形状可以在电压激励下改变。DE材料在未来的软致动器和传感器中具有很好的应用前景，例如软机器人，能量采集器和可穿戴传感器。在本文中，一个条纹DE致动器与集成传感能力的设计，制造和特点。由于条形致动器可以近似为柔性电容器，因此可以通过分析致动器的阻抗变化来检测致动器的位移。提出了一种在激励信号中加入高频探测信号的集成传感方案。利用快速傅立叶变换算法提取探测信号中电阻抗的变化，并采用人工神经网络非线性数据拟合方法检测执行器的位移。实验结果表明，通过改进数据处理和分析方法，该集成传感方法可以使测量误差控制在1%以内。
Dielectric elastomer (DE) is a type of soft actuating material, the shape of which can be changed under electrical voltage stimuli. DE materials have promising usage in future’s soft actuators and sensors, such as soft robotics, energy harvesters, and wearable sensors. In this paper, a stripe DE actuator with integrated sensing capability is designed, fabricated, and characterized. Since the strip actuator can be approximated as a compliant capacitor, it is possible to detect the actuator’s displacement by analyzing the actuator’s impedance change. An integrated sensing scheme that adds a high frequency probing signal into actuation signal is developed. Electrical impedance changes in the probing signal are extracted by fast Fourier transform algorithm, and nonlinear data fitting methods involving artificial neural network are implemented to detect the actuator’s displacement. A series of experiments show that by improving data processing and analyzing methods, the integrated sensing method can achieve error level of lower than 1%.