返回Google圖書搜尋

Artificial Intelligent Aided Terahertz Technology and Applications

出版	University of Alabama Libraries, 2023
URL	http://books.google.com.hk/books?id=VLs30AEACAAJ&hl=&source=gbs_api

註釋Terahertz (THz) technologies are at the forefront of emerging technologies for future applications, including bio-chemical sensing and imaging, non-destructive testing, biomedicine, security inspection, materials science, future 6G/7G communications, and more. Although terahertz techniques can be utilized in a wide range of areas, some limitations, such as high water absorption, limited spatial resolution, high costs, and complex analysis, prevent their widespread development. We purposed to develop and hasten the intricate design of THz metamaterials and analysis of THz spectra and image data in this research. At that point, combining Artificial Intelligence (AI) and THz technology was used to improve the generality and robustness of models that analyze and improve the performance of THz spectra and image data. We first developed THz spectroscopic images for biomedical applications such as distinguishing cancerous and healthy cells by independent component analysis-aided THz imaging. However, a characteristic of THz radiation, such as its strong water absorption, makes it challenging to reconstruct clear images of tissues in vivo or reconstructed THz images. By utilizing the ICA decomposition, THz spectroscopic images with high sensitivity of frequency dependence were achieved by significantly improving contrast and differentiation of the tumor region of the phantom cancerous cell. In the following study, combining artificial intelligence (AI) techniques with THz-TDS, we have demonstrated a method to obtain the conductivity of nanowire-based conducting thin films in a significantly effective, steady, and rapid manner. The training of neural networks has been simplified by utilizing time-domain waveforms rather than frequency-domain spectrums as input data. According to our neural network models, the calculated and predicted conductivity matched successfully.At last, the design of the stereo-metamaterial-based THz polarizers study was focused on in a rapid and efficient way. We actualized using the power of AI for rapid and high-efficiency inverse design of THz stereo-metamaterial polarizers and accelerated the analysis of the desired device. A tandem neural network (TNN) with a weighted loss function was successfully displayed to inversely design the SMM-based THz polarizer device from a desired ellipticity angle spectrum instead of the structure's reflection and phase spectra, to obtain the corresponding structural parameters for the first time.