The Research Of Human Body Pulse Signals Extraction And Classification

Pulse signal analysis of human body is a traditional Chinese medical diagnosis of the fundamental approach, but this method is now gradually disappearing only with a doctor's fingers feeling and experiences to judge. In order to improve the accuracy of pulse diagnosis, computer is used to denoise the human body pulse signals,, and to finish feature detection and recongnition.Due to the ability to reveal the signal details and local characteristics in the time-frequency domain, wavelet transformation is adopted in this paper. Wavelet packs decomposition is used to eliminate poer line interference, wavelet decomposition and reconstruction alogrithm is used to filter the power-line interference cause to respiration of human body, and then wavelet transformatin modelu maximum is used to detect character endpoint of denoised signal. The experimental results show that the pulse signal of human body denoised by wavelet transformation can effectively retain the pinnacle and mutation of pulse signals after eliminating the baseline wander and power line interference.The recognition system input uses different characters, including parameters from time and frequency domain analysis, harmoinc components of power spectrum, and spectral energy rate of different frequency components. In this paper, guassian function of membership grade from fuzzy control theory is introduced into neural network to form noraml fuzzy neural network, a kind of traditional Chinese medicine human body pulse signals recognition method based on normal fuzzy neural network is proposed finally. The experiment results show that, comparing with the traditional recognition methods, fuzzy neural network has higher recognition accuracy, faster training speed and global optimization. The average recognition rate using nine characters is 90.32%, and achieves better automatic classification performance of the pulse signal.