当前位置:主页 > 科技论文 > 自动化论文 >

基于支持向量机的无创血糖检测电极影响研究

发布时间:2019-04-24 18:14
【摘要】:伴随着生活水平的逐步提高,糖尿病成为了人群中的常见疾病,对预防糖尿病和一系列并发症的发生具有很重要的意义的是能及早地检测出血糖异常或监控血糖的日常变化,而有创检测或微创检测的检测方法使糖尿病患者的身心遭受到了巨大的痛苦。因此,无创血糖检测技术在自我检测方法中具有更好的应用前景。目前的无创血糖检测方法多种多样,由于电磁波法检测葡萄糖浓度的可靠性及稳定性,所以本文的研究以电磁波检测法为支撑。我国无创血糖检测技术研究起步较迟,主要是以大学科研形式存在。近年来,北京大学、清华大学、天津大学、西安交通大学、中国医科大学、中国科学院合肥智能机械研究所、第三军医大学等都开展了利用光学谱的无创血糖检测技术的基础研究。黄岚等人曾利用近红外漫反射光谱技术,采用傅立叶变换光谱仪进行无创定量检测血糖浓度的研究。此外莫希等人以光谱测量法为理论依据,建立了血糖无创检测的数理模型。国外利用电磁波进行无创血糖检测主要是采用微波频段进行检测。如日本Kokushikan大学的Y.Nikawa等人对微波检测法作了深入的研究,他们发现高于30 GHz的电磁波对人体组织来说,介电损耗表现为主要因素。然而,采用电磁波进行的无创血糖检测方法均未考虑电极尺寸和形状对检测结果的影响,所以本课题开展的主要工作为:1.分析目前全球糖尿病的患病现状及其危害,以及无创血糖检测技术的研究现状及成果,重点研究基于电磁波的无创血糖检测方法。2.无创血糖检测(基于电磁波的)国外现有的方法一般在高频段,可穿戴式设备大都与人体接触较频繁,则应使其工作频率在安全范围内,应选择合适的检测频段。3.国外基于电磁波的无创血糖研究选择S11或S21参数作为研究对象,本研究需对S参数做实验对比分析选取合适的S参数。4.现有的基于电磁波的无创血糖检测未考虑电极尺寸和形状对检测结果的影响,所以通过比较不同尺寸和不同形状的电极在模拟血糖溶液中的S参数,确定其对血糖检测的影响。5.利用libsvm算法对不同尺寸、形状电极的S参数值进行分类,验证实验结论的正确性。研究已取得进展:根据实验方案选定检测频段和S参数,并进行不同形状、尺寸电极的实验,分析实验数据表明电极形状、尺寸对S参数存在影响,且一定形状、大小的电极对S参数表现出一定规律,并对检测结果更理想。对实验所得的S参数进行SVM的分类算法验证,属于不同尺寸、形状电极的S参数值能被设计的分类器以较高正确率分类,验证了实验结果的正确性,即证明了电极的尺寸和形状对无创血糖检测存在影响,今后利用S参数进行无创血糖检测时设计合适形状、尺寸的电极,这对无创血糖检测精度的提高有一定的实际意义。
[Abstract]:With the gradual improvement of living standards, diabetes has become a common disease in the population, it is very important to prevent diabetes and a series of complications is to detect blood glucose abnormalities or monitor the daily changes of blood glucose as early as possible. Invasive or minimally invasive tests have caused enormous physical and mental pain in diabetic patients. Therefore, the non-invasive blood glucose detection technique has a better application prospect in the self-detection method. There are a variety of non-invasive blood glucose detection methods at present. Because of the reliability and stability of electromagnetic wave detection of glucose concentration, the research in this paper is supported by electromagnetic wave detection. The research on non-invasive blood glucose detection technology in China started late, mainly in the form of university scientific research. In recent years, Peking University, Tsinghua University, Tianjin University, Xi'an Jiaotong University, China Medical University, Hefei Institute of Intelligent Machinery, Chinese Academy of Sciences, The third military Medical University has carried out the basic research of non-invasive blood glucose detection technology using optical spectrum. Huang Lan and others used near-infrared diffuse reflectance spectroscopy to study the noninvasive quantitative detection of blood glucose concentration by Fourier transform spectrometer. In addition, based on spectral method, Moshi et al established a mathematical model for noninvasive detection of blood glucose. Non-invasive blood glucose detection by electromagnetic wave in foreign countries is mainly carried out in microwave frequency band. For example, Y.Nikawa of Kokushikan University in Japan has done in-depth research on microwave detection. They found that the dielectric loss of electromagnetic waves higher than 30 GHz is the main factor for human tissues. However, the non-invasive blood glucose detection methods using electromagnetic wave do not take into account the influence of electrode size and shape on the detection results, so the main work of this paper is as follows: 1. The present situation and harm of diabetes mellitus in the world and the research status and achievements of non-invasive blood glucose detection technology are analyzed, and the non-invasive blood glucose detection method based on electromagnetic wave is mainly studied. 2. Non-invasive blood glucose detection (based on electromagnetic wave) existing methods abroad generally in the high-frequency segment, wearable devices most of the more frequent contact with the human body, it should be made its working frequency within the safe range, should choose the appropriate detection band. 3. S11 or S21 parameters are selected as the research object in non-invasive blood sugar research based on electromagnetic wave in foreign countries. In this study, we need to do experimental comparative analysis on S parameters and select appropriate S parameters. 4. The existing non-invasive blood glucose detection based on electromagnetic wave does not take into account the influence of electrode size and shape on the detection results, so by comparing the S parameters of electrodes with different sizes and shapes in simulated glucose solution, Determine its effect on blood glucose detection. 5. The S parameters of different size and shape electrodes were classified by libsvm algorithm to verify the correctness of the experimental results. According to the experimental scheme, the detection frequency band and S parameters are selected, and the experiments of different shapes and sizes of electrodes are carried out. The analysis of experimental data shows that the shape and size of the electrode have influence on the S parameters, and the shape of the electrode has a certain shape. The size electrode shows a certain rule for S parameters, and the detection results are more satisfactory. The experimental S parameters are verified by SVM classification algorithm. The S parameter values of the shape electrode can be classified by the designed classifier with high accuracy, and the correctness of the experimental results is verified by the S parameter values of the shape electrode. That is to say, it is proved that the size and shape of the electrode have influence on the noninvasive blood glucose detection. In the future, the suitable shape and size of the electrode are designed when the S parameter is used to detect the noninvasive blood glucose, which is of practical significance to the improvement of the accuracy of the non-invasive blood glucose detection.
【学位授予单位】:成都理工大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:R587.1;TP18

【参考文献】

相关期刊论文 前2条

1 谷歆海;;网络分析仪的工作原理及在测量领域的应用[J];电子工程师;2008年07期

2 高荣强,范世福;现代近红外光谱分析技术的原理及应用[J];分析仪器;2002年03期

相关博士学位论文 前1条

1 卢宗武;基于脉搏波阻抗谱的血液电特性无创检测[D];天津大学;2007年

相关硕士学位论文 前3条

1 卢玉婷;基于SVM的闸机目标识别方法的研究[D];南京理工大学;2013年

2 付莉;面向对象的高分辨率遥感图像分类[D];电子科技大学;2012年

3 杜大莉;基于阻抗信息的乳腺组织BP网络识别方法[D];天津大学;2008年



本文编号:2464672

资料下载
论文发表

本文链接:https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2464672.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户80a97***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com