Research On Pulse Condition Signal’s Characteristics Based On Curve Fitting

Pulse wave signal contains a wealth of information about human physiology andpathology. Pulse detection and feature extraction can present health status of the humanand provide a possibility for the modern TCM pulse identification and classification. Thisis also the purpose of this article. Based on the pulse wave signal acquisition, signalfeature extraction and pulse recognition were analyzed and discussed.A comparative analysis was made about pulse research in recent years at home andabroad, and according to the characteristics of the pulse and actual conditions, a pulsesignal acquisition system was designed based on the piezoresistive sensor in order toachieve real-time acquisition of the pulse from the human radial artery and save the digitaldata as a text file into storage device.The wavelet threshold de-noising method based on wavelet theory was used to filterout the noise signal such as the body jitter, baseline drift and so on. On the basis of thepulse waveform structural features and its physiological significance, pulse waveform canbe superposed by the three main bell-shaped wave, so the least squares method based ontrust-region algorithm was used to realize the curve fitting of pulse data and finally ninecharacteristic parameters were achieved and as the basis of the pulse condition analysisand identification.Attribute partial sequence structure chart method based on formal concept analysistheory was used to take different pulse condition and their parameters as a formalbackground, and draw the corresponding attribute partial sequence structure chart. Someuseful information can be dug out and different pulse condition identification wereanalyzed and discussed. Besides, an assumption about the criteria for the classification ofpulse condition was proposed according to the analysis, which can provide a reference forpulse condition identification and classification.