Research And System Design Of Pulse Classification Algorithm Based On PPG Signal

With the rapid development of the times,people’s pace of life has gradually accelerated,and their attention to physical health has also decreased.Factors such as increased stress in life,insufficient sleep,and irregular diet have increased the incidence of cardiovascular diseases.Cardiovascular health is the core of human health.Increasing the popularity of cardiovascular disease knowledge,understanding cardiovascular related health issues in advance,and taking precautions can help reduce or even avoid the harm caused by cardiovascular disease.Pulse condition in traditional Chinese medicine are closely related to the health of blood vessels,which reflect the level of mobility of the heart,blood vessels,and internal organs throughout the body,as well as the health of the human body.Based on the above analysis,this paper proposes a pulse classification algorithm which takes the PPG signals as the research object,and designs and develops a pulse health detection intelligent system using the application software development platform aardio.The original PPG signal is collected using photoelectric sensors.Firstly,the Hilbert Huang transform(HHT)method is used to filter out the low-frequency baseline drift noise in the PPG signal,and then the motion artifact interference noise in the PPG signal is filtered based on the divergence value analysis to obtain a smooth and stable high-quality wave signal.The standard threshold reference range of the divergence value of the signal characteristics is calculated using the template signal that is not subject to the artifact interference.Then,the divergence value of the template signal and the experimental signal that is subject to the artifact interference is calculated,and the signal periods that have motion artifact interference in the experimental signal are identified,judged,and eliminated.The high-quality wave signal that is not subject to motion artifact interference is integrated,improving the reliability of the algorithm in wearable motion systems.The sliding window method combined with genetic algorithm is used to identify the peak and trough points of the main wave and the dicrotic wave of the pulse wave signal,and extract time-domain signal characteristics such as amplitude,slope,and area;Using Hilbert marginal spectrum analysis,the magnitude of the signal at various frequencies is obtained,and then the cumulative amplitude energy values at several frequency bands are obtained,which are used as time-frequency domain signal characteristics;Starting from the chaotic coefficient of the signal,the correlation dimension of the PPG signal time series is extracted as a feature of the nonlinear system signal,and the periodicity and degree of difference of the signal sequence are observed through phase space reconstruction of the pulse system.A feature series fusion algorithm is used to fuse signal features of time domain,time-frequency domain,and nonlinear systems to obtain a signal feature set that can comprehensively characterize the complete information of pulse classification categories;Combining filter and wrapper algorithms,a feature selection algorithm combining mRMR and SVM-RFE is used to filter the optimal subset of features with high correlation,low redundancy,and optimal classification performance of the classifier;Using artificial bee colony algorithm(ABC)to optimize the parameters of support vector machine(SVM),iteratively searching for the optimal parameter values of the penalty factor and kernel function parameters of the classifier.Through a large number of modeling studies and experimental comparisons,using artificial bee colony algorithm to optimize the classification model of SVM for pulse recognition and classification has the best effect.Finally,in order to facilitate physiological parameter testers to view their own pulse health status in real time,an intelligent pulse health detection system based on the application software development platform aardio and Python programming language was designed and developed to achieve the transformation of traditional Chinese medicine pulse diagnosis automation and visualization.The pulse health detection intelligent system includes a PPG signal acquisition module,a signal processing and data analysis module,and an application software programming and code writing module.System users can monitor their pulse health level in real time and seek medical attention in a timely manner when necessary.