Research And System Design Of Respiration Signal Extraction Algorithm Based On PPG

Respiratory signal,as one of the basic indicators for measuring human physiological health,reflects the coordinated operation of multiple physiological systems such as respiration,cardiovascular,and metabolism in the body.Directly measuring the respiratory signal in the traditional way can increase the difficulty for medical staff and also cause discomfort to patients.Therefore,in the rapidly developing field of intelligent healthcare,non-invasive and non-contact monitoring of respiratory rate has attracted increasing attention and research,and has significant implications for disease prevention and diagnosis of potential health problems.This thesis aims to propose an algorithm that can quickly and accurately extract respiratory signals from the Photoplethysmography(PPG)signal to improve the accuracy and efficiency of non-invasive respiratory monitoring,addressing the issues with traditional respiratory monitoring techniques.The algorithm combines the ensemble empirical mode decomposition(EEMD)and wavelet threshold denoising signal processing methods.Firstly,the PPG signal is decomposed using EEMD to obtain the intrinsic mode functions(IMFs)of the decomposed signal.Secondly,correlation analysis is performed on the IMF components,and the optimal IMF is selected through wavelet threshold denoising as the best component for reconstructing the respiratory signal.Finally,the reconstructed respiratory signal is compared and analyzed from three aspects: AR power spectral coefficient,relative coherence coefficient,and respiratory rate to verify the accuracy of the algorithm.A synchronous PPG signal and respiratory signal acquisition system based on the STM32F103 microcontroller was designed for the aforementioned algorithm.The system utilized the MAX30102 pulse oximeter sensor and the HKH-11 B piezoelectric respiratory sensor to acquire human PPG and respiratory signals.The hardware system included signal acquisition circuit modules,power supply modules,OLED display modules,and other circuits.The upper computer module,which established communication with the hardware using C++ language on the QT software platform,was used to display and save real-time pulse waveforms and respiratory waveforms.The EEMD-wavelet threshold algorithm was used to decompose the acquired data in MATLAB to achieve PPG signal decomposition and respiratory signal reconstruction.The final results showed a correlation coefficient above 0.92 for the AR power spectrum of the original and reconstructed respiratory signals,a waveform correlation coefficient above 0.7,and an accuracy rate above 0.95 for respiratory frequency.The research results of this thesis demonstrate the reliability of using the EEMD-wavelet threshold algorithm to extract respiratory signals from PPG signals,and provide a new approach for non-invasive monitoring in medical care.This has breakthrough significance for non-invasive wearable devices and clinical practice.