| As pulse signal measured on the blood vessel wall is considered as the response to the heart's excitation in the blood circulatory system, it brings the dynamic information of the whole system and can be used as the window of surveying system's feature. Researching the technology of extracting information from the pulse signal is not only as the technology basis of making progress of pulse diagnosis objectively, but also allows us to combine modern scientific and technological achievements with mysteriously ancient Chinese medicine theory so as to explore the mysteries of human beings. Therefore human pulse signal is chosen as the research subject in this paper, and the main research contents include:(1) Taking the human blood circulation system excited by the heart beat as the research object, and from the view of the energy transmission, the excitation, response and dynamic characteristics of human being's body as complex coupled system were analyzed based on modern testing technology theory. Then the strap-type integrated digital pulse sensor is chosen to achieve pulse signal acquainting.(2) With the feature of low entropy, multi-scale and multi-resolution, wavelet is used in data pre-processing. First the pulse signal is decomposed as multi-scale, so the effective signals and noise signals at different time scales show different frequency characteristics, which reveals the difference between them. Then the corresponding rules are constructed depending on these differences to realize threshold filtering process for the wavelet coefficients of signal and noise. Experiments show that comparing with the traditional Fourier analysis filtering, wavelet analysis effects are better in terms of noise suppression.(3) Pulse signals were transformed from time-domain into frequency domain based on FFT and Welch methods. The amplitude spectrum and power spectrum were taken as the subject and different types of pulse signals' frequency features were discussed combining with the pulse generation mechanism. Then it is programming achieved by means of Matlab for the intelligent extracting frequency domain parameters such as the main peak frequencies fi,spectrum energy percent SER (10) and PSER (10), bandwidth wij, energy percent p; and so on.(4) With the method of combining time domain analysis and with frequency domain analysis to establish time-frequency domain pulse-spectrum describing way for the human body system identification. Pulse-spectrum directly reveals the system's energy varying within a pulsation period, and related to the feature points on the pulse wave according to the vascular dynamics theory, has apparently corresponding energy peak. For the individual, different pulse cycles'pulse-spectrum has good agreement. For different human beings, there is obvious discrepancy between their pulse-spectrums when they are in different group and high similarity when they in same group and have similar physique. So the developed pulse-spectrum can be used as identifications of individual constitution and physiological state.(5) With Microsoft Visual Basic 6.0 for interface designing, MatLab 7.0 for the background of numerical computation and graphics rendering, and then the Dynamic Link Library technology for their integrating, a pulse signal information processing and extracting system is developed. Then the system is used in experimental research for data acquisition and processing, and samples'pulse signals were analyzed, by time-frequency methods. It makes signal acquisition, processing and extraction of information automated, which lays foundation for further large samples'experimental study. |
PDF Full Text Request