随着生活压力的增大，亚健康问题越来越严重，已经严重影响着人们的生活。目前，国内外对亚健康的评判主要是问卷调查的方式，这种方法存在主观随意性、且耗时。本文提出了一种脉搏信号与神经网络相结合的亚健康识别方法，首先提取脉搏信号的特征值，然后经过BP神经网络计算，实现对人体亚健康状态的客观识别。 本文的主要工作有： （1）用MATLAB编程实现脉搏信号的预处理和特征值的提取。这些特征值可用于训练和测试不同结构的BP神经网络。 （2）在ARM平台上实现对串口脉搏数据的读取与实时显示，并对读取的脉搏数据进行预处理和提取脉搏信号特征值。 （3）利用特征值训练BP神经网络，在QT平台上编写了亚健康识别应用程序。通过对脉搏信号特征值网络运算，实现了亚健康脉搏信号的分类识别。 （4）将亚健康脉搏识别程序移植到ARM平台中。 本文设计的亚健康脉搏识别系统，成功提取了脉搏信号的特征值，实现了对脉搏信号亚健康状态的识别。实验表明，该方法可用于对亚健康状态的检测。

With the increase of the pressure of life, the sub-health problem is more and more serious, which has seriously affected people's lives. At present, the domestic and international evaluation of sub-health is mainly a questionnaire survey, this method is subjective and time-consuming. In this thesis, a method of combining pulse signal with neural network is proposed to judge sub-health. First, the eigenvalue of pulse signal are extracted, and then calculated by BP neural network, to achieve the objective recognition of human sub-health status. The main work of this thesis is as follow: (1) Using MATLAB to realize preprocessing and eigenvalue extraction of pulse signal. These eigenvalues can be used to train and test BP neural networks with different structures. (2) On the ARM platform to achieve the serial pulse data reading and real-time display, and preprocess of the read pulse data and extraction of pulse signal eigenvalue. (3) The BP neural network is trained by the eigenvalue value, and the sub-health recognition application program is compiled on the QT platform. The classification and recognition of sub-health pulse signal is realized by the calculation of the eigenvalue value with the BP neural network. (4) Finally, the recognition program of sub-health is transplanted to the ARM platform. The sub-health pulse recognition system designed in this thesis has successfully extracted the eigenvalues of the pulse signal and realized the recognition of the sub-health state of the pulse signal. Experiments show that this method can be used to detect sub-health status.