| Heart rate is one of the important indicators reflecting the health of human heart.Heart rate monitoring not only reduces the accidents caused by cardiovascular disease,but also has great value for exercisers.In recent years,heart rate extraction technology based on wearable electrocardiograph(ECG)and photoplethysmography(PPG)has been rapidly developed and successfully applied to the field of human health monitoring in daily life.However,due to the limited computing resources of wearable hardware device and the strong noise interference in ECG and PPG signals,the human heart rate extraction of wearable device is not accurate.In order to solve the above problems,this paper proposes a full-state heart rate parameter extraction method of wearable device based on ECG / PPG.The main contents of this paper are as follows:1.In order to address the issue of inaccurate PPG heart rate extraction,imbalance of heart rate extraction accuracy and computational complexity caused by motion artifact noise in PPG heart rate extraction of wearable device,this paper proposes a heart rate parameter extraction algorithm of wearable device based on PPG frequency domain information.Firstly,This algorithm uses acceleration time domain information and PPG frequency domain information to evaluate the intensity of motion artifacts.Then,different heart rate estimation methods are selected according to different intensity of motion artifact.According to the order of motion artifact intensity from small to large,the frequency domain direct extraction method,the normalized LMS adaptive filtering method with momentum term and the improved joint sparse spectrum reconstruction method are used.Based on the performance characteristics of various methods,the overall framework of the algorithm is designed.Finally,experiment is conducted on the public data set,which verifies that the proposed algorithm can effectively reduce the computational complexity of the algorithm while maintaining a certain heart rate extraction accuracy.2.In order to solve the problem that the ECG signal noise is strong and irregular in the process of heart rate extraction of wearable ECG,which seriously affects the accurate extraction of heart rate parameters,this paper proposes a filter design method based on evolutionary learning in the signal preprocessing phase,which applies evolutionary learning to the filter design of R-peak detection algorithm.And the excellent global optimization ability of genetic algorithm is used to find the optimal filter.In the R-peak detection stage,a real-time adaptive threshold R-peak detection algorithm based on Brown exponential smoothing model is proposed.Based on the morphological characteristics and occurrence regularity of R-peak,the algorithm uses Brown exponential smoothing model to update the threshold parameters,and uses the relative error least square method to optimize the smoothing coefficient,so that the updated threshold can be more in line with R-peak detection and the accuracy of R-peak detection is improved.Finally,the proposed algorithm is tested on the self-built ECG data set,demonstrating its ability to adapt well to the strong noise environment and obtain satisfactory R-peak detection accuracy.3.In order to address the issue of inaccurate heart rate parameters caused by continuous severe noise interference in a single heart rate measurement method,which may lead to the loss of some key heart rate information,this paper proposes a heart rate parameter extraction algorithm based on ECG/PPG time-frequency domain information.By using the characteristics of PPG heart rate stability and all-day measurement and the characteristics of ECG heart rate real-time but susceptible to noise peak interference,the algorithm fuses ECG heart rate information and PPG heart rate information in the time dimension.It makes full use of their respective characteristics and advantages to form a consistent interpretation of cardiac pulsation,improve the accuracy of heart rate parameter extraction and strengthen the credibility of heart rate anomaly warning,so as to make up for the shortcomings of extracting heart rate parameters from a single signal source.Finally,the effectiveness of the algorithm is verified through experiments on a self-built data set. |
