Research On Weak Signal Detection And Analysis Approach In The Ballistocardiogram Information Monitoring System

With the improved demand for home monitoring, the synchronous detection of multi-physiological parameters becomes the research focus. In order to realize the health monitoring and disease early warning, we hope to achieve more human physiological state-informations through the convenient means when the subject doesn’t feel the detection. Ballistocardiogram information monitoring system could extract the BCG signal including multi-physiological information such as subject’s cardiovascular system function and breathing condition etc. with non-invasive situation, which completely meets the requirements of the development of daily monitoring. Through the review of the BCG signal research status, the method of detecting and analyzing weak physiological information from BCG signal is studied based on the sitting BCG signal. The main works are as follows:In the stage for analyzing BCG signal principle, build the mathematics model which is made up of the cardiac power generation and transmission in vitro and the detection outside the body. Then simulate the time domain waveform of this model with more cycles, and compare it with measured BCG signal to verify the accuracy of the simulation.In the stage for preprocessing BCG signal, analyze the property of BCG signal and make the preprocessing such as de-trending, smoothing filtering, and calculating autocorrelation function to purify the BCG signal deeply. At the same time, establish the simplified mathematics model of BCG signal to simulate the theoretical spectrum. And then the effective frequency range of the BCG signal is determined to filter the BCG signal and to make the wavelet denoising effectively.In the stage for detecting heart rate from BCG signal, propose a detecting method based on the chaos theory. Firstly, when the normal heart rate of subject is given, apply the Duffing chaotic oscillator to detect the weak periodic components from the BCG signal with lower SNR. In allusion to the output phase space trajectory, a kind of new criterion which is based on pulse coupled neural network is proposed to find the abnormal heart rate subject effectively from BCG signal. Secondly, when the normal heart rate of subject is unkown, propose an adaptive stochastic resonance method based on the linear random search algorithm, which could highlight the period component characteristics in the output waveform by changing the energy of noise to the energy of cardiac cycle composition to achieve the heart rate automatically. The experiments for this stage make subject’s ECG signal synchronously as assessment criteria to verify the accuracy of the algorithm. In the stage for detecting respiratory rate, propose three kinds of detecting methods to extract the respiratory rate from BCG signal. The first one is the adaptive interference cancellation algorithm, which achieve the respieratory component from the output by eliminating the cardiac cycle component between two channel BCG signals. The second one is the variable frequency complex demodulation algorithm, which demodulate the modulated signal--breathing composition by regarding the BCG signal as the result for respiratory component modulating cardiac cycle component. The third one is the envelope demodulation algorithm based on S transform, which extract BCG signal envelope from the angle of demodulating amplitude modulated signal. Lastly, detect the peaks from the respiratory component above to calculate respiratory rate. The experiments for this stage make subject’s nasal thermal signal synchronously as reference, compare with the wavelet approximate coefficients of BCG signal, and give the performance evaluation between these algorithms proposed.In the stage for extracting features of BCG signal, propose a blind segment mthod based on the diagnosis characteristics of BCG signal. Without the ECG signal as reference, the segment is realized and the time-domain feature point, frequency-domain characteristics, time-frequency singular value, and long and short term repeatability statistical characteristic parameters are extracted. Finally, propose two diagnose methods based on the deterministic feature points and the fuzziness cloud model respectively. The experiments for this stage apply the arrhythmia subjects to verify the superiority of the proposed methods.In the last, design and realize a set of ballistocardiogram information monitoring system. This system makes the sitting BCG signal detecting chair designed by laboratory as the hardware device to provide sufficient BCG data for the experiments above. At the same time, make the automatic analysis platform of BCG signal based on LabVIEW as the software environment. And test the correctness of the method proposed and the practical utility of this system through the detecting process of arrhythmia subjects.