电子器件2024，Vol.47Issue(1)：255-264,10.DOI:10.3969/j.issn.1005-9490.2024.01.042
基于近红外光谱法和FPGA的纺织物成分检测系统
Textile Component Detection System Based on Near-Infrared Spectroscopy and FPGA
摘要
Abstract
Quantitative determination of fiber composition is a crucial aspect of textile testing.Meeting the market's demand for accu-rate,rapid,and eco-friendly fiber composition analysis requires effective testing methods and instruments.A textile composition detection system that utilizes NIR spectroscopy and FPGA technology is presented.The system captures NIR spectral data using a Hamamatsu MEMS-FPI sensor,employs a one-dimensional convolutional neural network for modeling,reduces network complexity through knowledge distillation,and finally deploys the network on an FPGA.Experimental results demonstrate the effectiveness of the proposed approach with a strong detection performance(R-squared>0.82).Additionally,the system achieves favorable outcomes with low power consumption and minimal latency(<1 ms).关键词
纺织物成分检测/卷积神经网络/近红外光谱法/FPGA/知识蒸馏Key words
textile composition determination/convolution neural network/near-infrared spectroscopy/FPGA/knowledge distilling分类
信息技术与安全科学引用本文复制引用
张恒瑞,姚丞韬,刘向..基于近红外光谱法和FPGA的纺织物成分检测系统[J].电子器件,2024,47(1):255-264,10.基金项目
国家重点研发计划项目(2022YFB4401300) （2022YFB4401300）
苏州市姑苏创业领军人才项目(ZXL2021303) （ZXL2021303）
