As a key physiological signal,electrocardiogram(ECG)is widely used in the medical field.However,the ECG signal is easily interfered by noise during the collection process,which affects the signal quality.An improved algorithm of singular spectrum analysis(SSA)is designed for ECG signal noise reduction.The logistic regression(LR)algorithm is introduced in the principle component grouping stage of SSA,and the principle component method is improved to automatic grouping to realize SSA self-supervised noise reduction processing for ECG signals.Using an ECG signal acquisition device based on AD620,of 53 ECG signals was constructed as a testing set for verification.Using the improved algorithm of singular spectrum analysis,the accuracy of automatic selection of principal elements was 98.68%. The signal-to-noise ratio(SNR)of the reconstructed ECG signal increased from 10.43 dB to 20.17 dB on average.It can extract clear PQRST waves effectively,and has good practical prospects in ECG signal detection and noise reduction in the medical field.