当前位置: X-MOL 学术Appl. Mathmat. Model. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Deep learning for simultaneous measurements of pressure and temperature using arch resonators
Applied Mathematical Modelling ( IF 4.4 ) Pub Date : 2021-01-11 , DOI: 10.1016/j.apm.2021.01.006
Mehdi Ghommem , Vladimir Puzyrev , Fehmi Najar

The ability to measure pressure and temperature using a MEMS sensor constitutes a major interest for several engineering applications. In this paper, we present a method and system for simultaneous measurements of pressure and temperature using electrically-actuated arch resonators. The sensor design is selected so that the arch microbeam is sensitive to temperature variations of the surrounding via the inherent thermal stress and to pressure change via the squeeze-film damping resulting from the air flow between the microbeam and the fixed underneath electrode (substrate). A physics-based model is formulated and validated by comparing the static deflection of the microbeam and its natural frequencies under varying temperature to experimental data reported in the literature. We use deep learning to estimate the pressure and temperature from the natural frequencies, quality factors and static deflection of the microbeam. Results show accurate prediction of the temperature and pressure from the quality factors of the arch resonator based on the first three vibration modes. Further improvement is achieved by adding the natural frequencies to the input data. The robustness of the deep learning approach to noise is demonstrated by the small errors obtained using different loss functions when introducing different noise levels to the training data. The proposed approach allows, for the first time, the combination of arch beams dynamics and deep learning techniques for simultaneous sensing of pressure and temperature.



中文翻译:

深度学习使用拱形谐振器同时测量压力和温度

使用MEMS传感器测量压力和温度的能力成为一些工程应用的主要兴趣。在本文中,我们介绍了一种使用电控拱形谐振器同时测量压力和温度的方法和系统。传感器的设计经过选择,以使弓形微束通过固有的热应力对周围的温度变化敏感,并通过由微束和固定的下部电极(基板)之间的气流产生的挤压膜阻尼对压力变化敏感。通过将微束的静态挠度及其在不同温度下的固有频率与文献中报道的实验数据进行比较,来制定和验证基于物理学的模型。我们使用深度学习从微束的自然频率,品质因数和静态挠度估算压力和温度。结果表明,根据前三个振动模式,可以根据拱形谐振器的品质因数准确预测温度和压力。通过将固有频率添加到输入数据中,可以实现进一步的改进。当将不同的噪声水平引入训练数据时,使用不同的损失函数获得的小误差证明了深度学习方法对噪声的鲁棒性。所提出的方法首次允许将拱梁动力学和深度学习技术相结合,以同时感测压力和温度。结果表明,根据前三个振动模式,可以根据拱形谐振器的品质因数准确预测温度和压力。通过将固有频率添加到输入数据中,可以实现进一步的改进。当将不同的噪声水平引入训练数据时,使用不同的损失函数获得的小误差证明了深度学习方法对噪声的鲁棒性。所提出的方法首次允许将拱梁动力学和深度学习技术相结合,以同时感测压力和温度。结果表明,根据前三个振动模式,可以根据拱形谐振器的品质因数准确预测温度和压力。通过将固有频率添加到输入数据中,可以实现进一步的改进。当将不同的噪声水平引入训练数据时,使用不同的损失函数获得的小误差证明了深度学习方法对噪声的鲁棒性。所提出的方法首次允许将拱梁动力学和深度学习技术相结合,以同时感测压力和温度。当将不同的噪声水平引入训练数据时,使用不同的损失函数获得的小误差证明了深度学习方法对噪声的鲁棒性。所提出的方法首次允许将拱梁动力学和深度学习技术相结合,以同时感测压力和温度。当将不同的噪声水平引入训练数据时,使用不同的损失函数获得的小误差证明了深度学习方法对噪声的鲁棒性。所提出的方法首次允许将拱梁动力学和深度学习技术相结合,以同时感测压力和温度。

更新日期:2021-01-19
down
wechat
bug