Study Of FECG Signal Processing Based On Model-Free Adaptive Control And Non-negative Blind Separation

FECG can reflect the growth and health of the fetus in the mother.Analysis of FECG assessment is an important means of monitoring the fetus.Since the fetal electrocardiogram needs to pass through the amniotic fluid and the mother's abdomen tissue,It is very weak and easily overwhelmed by the mother's ECG and other noise.FECG is a low-frequency and random non-stationary signal.Therefore,Extracting FECG accurately has always been a difficulty and hot spot for bio-signal processing.By studying deeply excellent research,this paper carried out a research on FECG extraction based on model-free adaptive control and non-negative blind separation.Firstly,Make an pre-process on the collected mixed signal,design comb filter to remove making frequency interference,low-pass filter to remove high-frequency noise,and modelfree remove baseline drift adaptively.Then,a blind separation method based on kurtosis and skewness is introduced.According to the difference between the absolute value of the eccentricity and skewness of the fetal ECG signal and the difference between absolute value of the eccentricity and skewness of the mother's ECG signal,An objective function based on third-order and fourth-order statistics is used to extract the mother ECG signal and the FECG signal with noise.Finally,a non-negative blind separation algorithm is proposed to extract the pure FECG signal from the noisy FECG signal.Non-negative blind separation is Is a method of extracting low-dimensional structure from high-dimensional space through non-negative constraints.The FECG signal containing noise is equivalent to the low-dimensional structure.First,It is time-frequency changed for fetal ECG which contains noise,and then the pure fetal ECG signal is extracted by the non-negative blind separation algorithm.The algorithm is verified validity by simulation experiments and results analysis.Firstly,use known analog signals to verify the reliability of the model-free adaptive algorithm on removing baseline drift.The comparison between the baseline and the wavelet transform is used to verify the superiority of the algorithm.Secondly,Make the simulation on the data by the algorithm of extracting fetal ECG by non-negative blind separation,the three-way ECG signal is compared by the correlation analysis and the non-negative blind separation effect to obtain the optimal one-way signal,after R wave detection,The accuracy of the algorithm is verified by the comparison between instantaneous heart rate calculation and the original signal.Finally,The experiments on the actual data are used to verify the stability and extensiveness of the algorithm.In order to Promote the above algorithm to practical application,a fetal heart electric extraction system is designed and introduced.Collect the abdominal wall signal by device and extract the FECG signal by the above algorithm,and display the real-time monitoring on the page,so that people who know a little electric knowledge can diagnose the fetal health status at any time.