Research On Extraction Algorithm Of Blood Oxygen Saturation For Motion Process

Oxygen saturation?SpO₂?refers to the percentage of oxyhemoglobin?HbO₂?bound by oxygen molecules in the blood to the total bound hemoglobin?Hb?content.It can also be expressed as oxygen concentration in the blood.It is an important physiological parameter that characterizes the health of the human respiratory system.However,due to noise interference during exercise,the measured blood oxygen saturation tends to have a large error.For this reason,this paper analyzes the source and characteristics of noise,and proposes a blood oxygen saturation extraction algorithm oriented to the motion process.When the human body moves,because the photoplethysmography?PPG?signal is affected by Motion Artifact?MA?and vasodilation and other factors,how to extract the correct blood oxygen saturation from it has always been a research problem..After investigation and implementation of several different filtering algorithms,using discrete wavelet transform,the photoplethysmography pulse signal can be decomposed into wavelet coefficients of different levels of frequency,and then the empirical mode decomposition is applied to the wavelet coefficients containing pulse signal components.?Empirical Mode Decomposition,EMD?method to remove motion artifacts in the pulse wave signal;for non-stationary signals,adaptive filter algorithm is used,its advantage is that it can adaptively filter the filter according to changes in the external environment.The parameters and structure are adjusted accordingly,and the error between the filtered signal and the desired signal is minimized through continuous iteration.The adaptive filter is affected by the convergence speed and the steady-state error.The use of a variable step-length least-mean-square algorithm can effectively improve the calculation accuracy,and uses the acceleration signal collected by the Bosch BMA250 chip as a reference signal.For the de-noised pulse wave signal,the mathematical morphology method was used to extract the peaks and troughs.The peak-valley extraction operator was constructed by basic morphological operations,and the peaks and troughs of the pulse wave were accurately extracted.In order to verify the accuracy of the algorithm proposed in this paper,we used static and motion experiments.We collected data of 5 subjects under static and motion conditions.Through three experiments,a total of 30 sets of data were collected.The two denoising algorithms proposed in this paper were used to denoise the data.At the same time,the commonly used dip oximeters on the market were used as controls.According to the Lambert-Beer law,the formula for blood oxygen saturation calculation was deduced,and then the parameters of the calculation formula were calibrated by the blood oxygen simulator and applied to the comparative test.By analyzing the statistical indicators such as the average absolute error and the average absolute error percentage,it is verified that the proposed algorithm based on discrete wavelet transform and empirical mode decomposition has higher accuracy than the adaptive filter based on minimum mean square error.The results show that the algorithm based on the combination of discrete wavelet transform and empirical mode decomposition has a higher accuracy in the calculation of oxygen saturation for sports anti-interference.