Identify Sudden Cardiac Death Signals Based On Cyclostationary Characteristics Of ECG Signal

Sudden cardiac death is a kind of disease which cause se rious damage to the human life,most scholars will limit in the sudden death of time 1 hour.If we can warn before the sudden cardiac death happen,the patient's life can be saved.The ECG signal is a reflection of the mechanical systolic and diastolic heart,it is not a stationary signal.Based on the cyclostationary features of ECG signal,when people taking place sudden cardiac death in the body,the cyclostationary features of ECG signal will also change.The ECG signal is not stationary,but traditional identification of cardiovascular disease is aiming at stationary signal.So a new method is proposed to extract the cyclostationary characteristics of ECG signal.Support vector machine is used to identify the sudden cardiac death The following several aspects of the research work have mainly been completed in this thesis:(1)ECG signal filtering is the precondition of feature extraction.The common noise in the process of ECG signal acquisition is analyzed.Through the suitable for not stationary signal processing of the wavelet transform filtering out the interferences of ECG signal.At first,to model ECG signal which can produce clean signal,through adding different signal to noise ratio,to evaluate the effect of the filters;second,wavelet transform filter is designed to filter these interferences.Compared with the coefficient of filter,the wavelet transform of ECG signal filtering effect is better.Finally with actual ECG signal validated the wavelet transform,the experimental results show that wavelet transform filter can effectively remove high frequency and low frequency interference.(2)Aiming at the cyclostationary features of ECG signal,firstly the basic concepts of first-order and second-order cyclic stationary are introduced.Reflecting the cyclostationary features of ECG signal formula integral loop power spectral density function was deduced;Then analyzed the real-time high temporal smoothing cyclic spectrum estimation algorithm——Fast Fourier Transform Accumulation Method.To estimate the cycle spectrum of sine signal.The results show that cyclic spectrum simulation estimation results are consistent with the theoretical calculation results;finally,to extract the different typical crowd circulation steady characteristics of ECG signal.According to the characteristics of the interferences in ECG signals,the principle of cyclostationary features of ECG signal is used to detect interferences.(3)Aiming at sudden cardiac death diseases identification accuracy is not high,based on the cyclostationary features of ECG signal,Adopt the method of cyclic spectrum estimation in cyclic frequency domain to extract different typical crowd of cyclostationary characteristics.Support vector machine is used to identify the sudden cardiac death.Compared the two classes of linear classifier and support vector machine classification effect,effect of support vector machine is better.Through contrast with sudden cardiac death existing recognition methods.The result show that cyclic frequency average can especially reflect the cyclostationary characteristics of ECG signal and accurately identify the sudden cardiac death.Sudden death of ECG signal recognition accuracy up to 97.50%.