本文以壓電換能器 (PZT) 擔任超聲波發送與接收系統為基礎，在薄板中產生蘭姆波 (Lamb wave)，以手指觸碰的波型變化為數據，結合人工智慧 (Artificial Intelligence)，完成以壓電/超音波為基礎的觸控感測器。
本研究以不鏽鋼和玻璃為觸控面板材料，並以多通道的資料擷取 (Data Acquisition, DAQ) 卡負責電壓的輸出與接收，本研究利用PZT不同的擺放方式激振出A-mode和S-mode的蘭姆波，壓電感測器環狀式和直線式排列觸控面板以A-mode蘭姆波為基本信號，玻璃觸控面板則是以S-mode蘭姆波為基本信號，結合本實驗室開發的自動擷取系統搭配假手指收取訊號，將此信號使用卷積神經網路 (Convolution Neural Networks, CNN) 訓練辨識模型，再進行即時觸控辨識，以得到觸控位置。
本研究完成具備人工智慧判讀能力的超音波觸控感測面板，首先，以真手指進行實測之後發現在壓電感測器環狀式架構可以達成較大感測面積160×160 mm2、觸碰準確度大約為92%、觸控解析度為1×1 cm2 、反應時間約為60 ~ 65 ms 左右，且受溫度的影響不大；其次，在壓電感測器直線式排列下，發現在相同邊界反射情況下比較所收取訊號時間長短的準確度以擷取時間較長的準確度較高，另外，在較小邊界反射架構下，能達成準確率95% 且解析度為5×5 mm2以及反應時間約為20 ~ 25 ms，顯示CNN演算法對邊界條件尺寸的改變並不受影響，最後玻璃板則是因多通道資料擷取系統的硬體限制導致無法達成觸控面板的目標，不過在訊號測量上可以接收到所期望的訊號，顯示此架構未來發展的可行性。

This thesis applies piezoelectric transducers to generate Lamb waves on a thin plate for establishing an ultrasound sending and receiving system. Based on analyzing the change of acoustic signals induced by finger touching on the plate, an ultrasound-based touch panel is fulfilled. To enhance the positioning accuracy and sensitivity of this touch panel, convolution neural network (CNN) in artificial intelligence (AI) is adopted here and integrated with the Lamb wave ultrasound system.
In this work, stainless steel and glass plate are used as touch panel materials, and multi-channel data acquisition (DAQ) card is used for exciting and recording the voltage signals from piezoelectric transducers. Both A-mode and S-mode Lamb waves are excited by different piezoelectric transducer deployment configurations. The ring-shaped and linear-shaped arrangements are mainly for the A-mode Lamb waves in the stainless-steel plates, while for the glass plate S-mode Lamb waves are utilized using piezoelectric transducers attached on the side walls of the plate. After establishing the ultrasound system for monitoring the Lamb waves propagating in the plate, an artificial finger carried by an x-y-z automatic stage is used for creating and collecting numerous wave signals, which are then used for training a CNN model. Finally, the ultrasound system and the CNN model can be applied for real-time recognition and positioning of human’s finger touch.
Experiments have been carried out on the Lamb wave-based touch screen. It is found that a larger sensing area of 160×160 mm2 and a touch resolution of 1×1 cm2, touch accuracy can be achieved about 92% and the response time is around 60 ~ 65 ms under the ring-shaped arrangement in steel plate. It also shows that this performance is not seriously affected by temperature. Secondly, for the linear-shaped arrangement in steel plate, the touching accuracy is about 95 % with a touch resolution of 5×5 mm2 and a response time of about 20 ~ 25 ms. It is also shown that the CNN algorithm is not affected by the reverberating waves of different boundary size. Finally, the glass plate is unable to achieve the goal of the touch panel due to the hardware limitations of the multi-channel DAQ system. But the desired signal can be received which shows the feasibility of the development of this kind of architecture in the future.

[1] A. Jain, D. Bhargava, and A. Rajput, “Touch-Screen Technology,” International Journal of Advanced Research in Computer Science and Electronics Engineering, vol. 2, no. 1, pp. 74-78, 2013.
[2] Y. Liu, J. P. Nikolovski, N. Mechbal, M. Hafez, and M. Vergé, “A Multi-Touch Plate Based on Lamb Wave Absorption,” Procedia Chemistry, vol. 1, no. 1, pp. 156-159, 2009.
[3] Y. Liu, J. P. Nikolovski, N. Mechbal, M. Hafez, and M. Vergé, “An Acoustic Multi-Touch Sensing Method Using Amplitude Disturbed Ultrasonic Wave Diffraction Patterns,” Sensors and Actuators A: Physical, vol. 162, no. 2, pp. 394-399, 2010.
[4] Y.-T. Lin and Y.-C. Lee, “Application of Piezoelectric Actuator/Sensor and Lamb Wave for Touch Recognition,” M.S. Thesis, Department of Mechanical Engineering, National Cheng Kung University, Tainan, Taiwan, 2017.
[5] R. Adler and P. J. Desmares, “An Economical Touch Panel Using SAW Absorption,” IEEE Transactions on Ultrasonics, Ferroelectrics, and Electronic Engineering, vol. 34, no. 2, pp. 195-201, 1987.
[6] R. Adler and P. J. Desmares, “SAW Touch Systems on Spherically Curved Panels,” IEEE 1986 Ultrasonics Symposiums, Williamsburg, Virginia, USA, Nov. 17-19, 1986.
[7] Y. Tanaka, M. Takeuchi, J. L. Aroyan, and J. Kent, “P2I-2 Characteristics of SAW Grating Reflectors with Arbitrary Angels of Indices in Ultrasonic Touch Sensors,” 2006 IEEE Ultrasonics Symposiums, Vancouver, British Columbia, Canada, Oct. 2-6, 2006.
[8] K. Firouzi, A. Nikoozadeh, and B. T. Khuri-Yakub, “Numerical Modeling of Ultrasonic Touchscreen,” IEEE International Ultrasonics Symposium Proceedings, Chicago, Illinois, USA, Sept. 3-6, 2014.
[9] K. Firouzi, A. Nikoozadeh, T. E. Carver, and B.T. khuri-Yakub, “Lamb Wave Multitouch Ultrasonic Touchscreen,” IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 63, no. 12, pp. 2174-2186, 2016.
[10] A. Raghavan and Carlos E. S. Cesnik, “3-D Elasticity-Based Modeling of Anisotropic Piezocomposite Transducers for Guided Wave Structural Health Monitoring,” ASME Journal of Vibration and Acoustic, vol. 129, no. 6, pp. 739-751, 2007.
[11] A. Rahavan and Carlos E. S. Cesnik, “Modeling of Piezoelectric-Based Lamb Wave Generation and Sensing for Structural Health Monitoring,” Smart Structures and Materials, San Diego, California, Jul. 29, 2004.
[12] Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-Based Learning Applied to Document Recognition,” Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, 1998.
[13] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, “Learning Internal Representation by Error Propagation,” Parallel Distributed Proceeding: Explorations in the Microstructure of Cognition: Foundations, MIT Press, pp. 318-362, 1987.
[14] P. Y. Simard, D. Steinkraus, and J. C. Platt, “Best Practices for Convolutional Neural Networks Applied to Visual Document Analysis,” Seventh International Conference on Document Analysis and Recognition, Edinburgh, UK, Aug. 6, 2003.
[15] N. Jmour, S. Zayen, and A. Abdelkrim, “Convolutional Neural Networks for Image Classification,” IEEE Internal Conference on Advanced System and Electric Technologies, Hammamet, Tunisia, Mar. 22-25, 2018.
[16] C.-S. Chang, Y.-C. Lee, and Y.-C. Wang, “An Ultrasonic Touchscreen Sensors Based on Piezoelectric Transducers and Artificial Intelligence,” M.S. Thesis, Department of Mechanical Engineering, National Chung Kung University, Tainan, Taiwan, 2019.
[17] Y.-T. Chen, Y.-C. Lee, and Y.-C. Wang, “Machine Learning Applied to Ultrasonic Touch,” M.S. Thesis, Department of Mechanical Engineering, National Chung Kung University, Tainan, Taiwan, 2019.
[18] W. S. McCulloch and W. Pitts, “A Logical Calculus of The Ideas Immanent in Nervous Activity,” Bulletin of Mathematical Biophysics, vol. 5, no. 4, pp. 115-133, 1943.
[19] Y. Lecun, C. Cortes, and Christopher J.C. Burges, “THE MNIST Database of Handwritten Digits,” Retrieved from http://yann.lecun.com/exbd/mnist/.
[20] V. Ewald, R. M. Groves, and R. Benedictus, “DeepSHM: A Deep Learning Approach for Structural Health Monitoring Based on Guided Lamb Wave Techniques,” SPIE Smart Structure + Nondestructive Evaluation, Denver, Colorado, USA, Mar. 3-7, 2019.
[21] Pieter-Tjerk de Boer, D. P. Kroese, S. Mannor, and R. Y. Rubinstein, “A Tutorial on the Cross-Entropy Method,” Annals of Operations Research, vol. 134, no.1, pp. 19-67, 2005.
[22] D. P. Kingma, J. Ba, “Adam: A method for Stochastic Optimization,” 3rd International Conference on Learning Representations, San Diego, California, May. 7-9, 2015.
[23] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolution Neural Networks,” Communications of the ACM, vol. 60, no. 6, pp.84-97, 2017.