穿戴式脉象检测装置及其信号处理的研究
发布时间：2018-09-17 20:23
【摘要】：中医脉诊学是中国传统文化的重要组成部分,是中华文明的艺术瑰宝。中医博大精深,源远流长,几千年来中医学为中华民族的繁衍健康做出了重要贡献,至今仍是医疗实践中一门不可或缺的重要学科。随着社会科学的发展,脉象仪作为中医与现代科技结合的结晶,解决了中医脉诊的客观化和科学化的问题,然而现阶段研制的脉诊仪多停留在实验室阶段,难以推广和应用,并且随着社会老龄化,人们对医疗穿戴式设备的需求变得十分迫切。为了应对上述状况,本文设计了一种基于中医脉诊的穿戴式脉象检测和分析系统。硬件方面通过涡轮蜗杆及螺纹传动结构解决装置的尺寸和传动问题,采用贴合中医采脉手法的压阻式传感器来提取脉象信号,并设计了自动寻找最佳取脉压力算法,去模拟中医浮、中、沉的采脉手法；另外结合传感器的输出信号的特点设计了对应的信号调理电路；软件方面下位机控制采用arduino控制器,通过蓝牙与上位机进行通讯,上位机手机端实现信息的采集、波形的绘制、时域特征的提取及下位机的控制交互等功能。脉象分析方面,采用非线性分析方法中的递归定量法分析了平脉、滑脉、弦脉三种脉象信号,从构建的相空间重构图可以发现脉象信号有明显周期性。从精确递归图中提取了脉象信号的10个参数,随后利用灰度共生矩阵对饱和递归图提取了纹理特征的4种参数。本文对平脉、滑脉、弦脉提取的参数进行了两两对比并分析了脉象之间的差异。最后,利用深度信念网路(DBN)的方法对三种脉象信号进行分类,通过调节不同隐层数和节点,分析对分类准确率的影响,并最终确定合适的隐层数和节点。
[Abstract]:Pulse diagnosis is an important part of Chinese traditional culture and an artistic treasure of Chinese civilization. Traditional Chinese medicine (TCM) has a long history. It has made important contributions to the health of the Chinese nation for thousands of years and is still an indispensable subject in medical practice. With the development of social science, pulse instrument, as a combination of traditional Chinese medicine and modern science and technology, has solved the problem of objectification and scientization of pulse diagnosis of traditional Chinese medicine. However, the pulse diagnosis instrument developed at this stage is mostly in the laboratory stage, so it is difficult to popularize and apply it. And with the aging of society, people's demand for medical wearable devices becomes very urgent. In order to deal with the above situation, a wearable pulse detection and analysis system based on TCM pulse diagnosis is designed. On the hardware side, through the turbine worm and screw drive structure to solve the problem of the size and transmission of the device, the piezoresistive sensor with traditional Chinese medicine pulse collecting technique is used to extract the pulse signal, and the automatic algorithm for finding the best pulse pressure is designed. In addition, the corresponding signal conditioning circuit is designed in combination with the characteristics of the output signal of the sensor. In software, the lower computer control adopts arduino controller and communicates with the upper computer through Bluetooth. The functions of information collection, waveform drawing, time domain feature extraction and the control interaction of the lower computer are realized on the mobile terminal of the upper computer. In the aspect of pulse analysis, three kinds of pulse signals, flat pulse, smooth pulse and chord pulse, are analyzed by recursive quantitative method in nonlinear analysis method. From the constructed phase space reconstruction diagram, it can be found that the pulse signal has obvious periodicity. Ten parameters of pulse signal are extracted from accurate recursive image, and then four parameters of texture feature are extracted by gray level co-occurrence matrix. In this paper, the extraction parameters of flat, smooth and chord veins are compared and the differences between pulse patterns are analyzed. Finally, three pulse signals are classified by using the method of deep belief network (DBN). By adjusting the number of hidden layers and nodes, the effect on classification accuracy is analyzed, and the appropriate hidden layers and nodes are finally determined.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN911.7;TH789
本文编号：2246988
[Abstract]:Pulse diagnosis is an important part of Chinese traditional culture and an artistic treasure of Chinese civilization. Traditional Chinese medicine (TCM) has a long history. It has made important contributions to the health of the Chinese nation for thousands of years and is still an indispensable subject in medical practice. With the development of social science, pulse instrument, as a combination of traditional Chinese medicine and modern science and technology, has solved the problem of objectification and scientization of pulse diagnosis of traditional Chinese medicine. However, the pulse diagnosis instrument developed at this stage is mostly in the laboratory stage, so it is difficult to popularize and apply it. And with the aging of society, people's demand for medical wearable devices becomes very urgent. In order to deal with the above situation, a wearable pulse detection and analysis system based on TCM pulse diagnosis is designed. On the hardware side, through the turbine worm and screw drive structure to solve the problem of the size and transmission of the device, the piezoresistive sensor with traditional Chinese medicine pulse collecting technique is used to extract the pulse signal, and the automatic algorithm for finding the best pulse pressure is designed. In addition, the corresponding signal conditioning circuit is designed in combination with the characteristics of the output signal of the sensor. In software, the lower computer control adopts arduino controller and communicates with the upper computer through Bluetooth. The functions of information collection, waveform drawing, time domain feature extraction and the control interaction of the lower computer are realized on the mobile terminal of the upper computer. In the aspect of pulse analysis, three kinds of pulse signals, flat pulse, smooth pulse and chord pulse, are analyzed by recursive quantitative method in nonlinear analysis method. From the constructed phase space reconstruction diagram, it can be found that the pulse signal has obvious periodicity. Ten parameters of pulse signal are extracted from accurate recursive image, and then four parameters of texture feature are extracted by gray level co-occurrence matrix. In this paper, the extraction parameters of flat, smooth and chord veins are compared and the differences between pulse patterns are analyzed. Finally, three pulse signals are classified by using the method of deep belief network (DBN). By adjusting the number of hidden layers and nodes, the effect on classification accuracy is analyzed, and the appropriate hidden layers and nodes are finally determined.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN911.7;TH789
【参考文献】
相关期刊论文 前1条
1 周时伟;谢维波;;基于Android的智能家居终端设计与实现[J];微型机与应用;2012年14期
，本文编号：2246988
本文链接：https://www.wllwen.com/kejilunwen/yiqiyibiao/2246988.html
