Research On Human Multiphysiological Parameters Monitoring Fusion Motion State Recognition

Nowadays,various portable smart mobile terminal devices have received wide use,and with the types of sensors in smart devices sharply increasing,achieving convenient monitoring of physiological parameters is necessary.Through identifying and analyzing different motion states,the real physiological information contained in monitoring data is better excavated.This paper focuses on the changes of physiological parameters during human movement.The main work of this paper is concluded as follows:1.The results of collecting six motions data of 12 experimenters by single accelerometer show that the best acceleration data acquisition position is on lumbar median.A motion recognition method based on multiclassifier fusion is proposed to overcome the problem that single classifier has different recognition effects on different motion,which significantly improve the overall recognition effect.2.An error correction algorithm of heart rate detection based on GBDT is proposed to correct the error of heart rate monitoring by ECG signals in motion.Analyzing 11 experimenters' motion data by 5-fold crossvalidation shows that the algorithm has strong robustness and significant correction effect,and the average error of heart rate detection is reduced from 10.60 bpm to 5.76 bpm,with an average increase of 45.70%.3.With the studying of time and frequency domain analysis methods of heart rate variability,a variable step-size firefly algorithm is proposed to optimize the parameters of SVM to judge abnormal heart rate and blood pressure parameters.The analysis of MIMIC data samples by 10-fold cross-validation shows that the average recognition accuracy is increased from 90.6% to 93.5%.4.The application requirements of multiple physiological parameters based on motion recognition are investigated by network questionnaire,three groups of experiments are proposed according to the investigation results,including collecting 24 experimenters to study the changes of heart rate,blood pressure and body temperature,12 experimenters to analyze the validity of fusing physiological parameters to improve the recognition rate of motion and 14 experimenters to discuss the health status of physiological parameters introduced with motion interference.The experimental results show that motion recognition with physiological features can effectively improve the recognition rate of motion,with the average recognition rate rising from 94.8% to 95.7%.The recognition accuracy of SVM-based blood pressure and heart rate abnormality detection algorithm is over 87%.