Denoising of ECG Signals Based on CEEMDAN

Heart disease is one of the major diseases of human health, and ECG can reflect the health condition of the heart to a certain extent. In order to reduce the noise in ECG signals, this paper proposes a CEEMDAN based, i.e., adaptive noise complete empirical mode decomposition and state machine logic method to extract feature waveforms from ECG signals. First, the noise-containing ECG signal is CEEMDAN decomposed to obtain 10 IMF components and one residual component, and the low-frequency IMF component, i.e., the baseline drift signal, is determined using the over-zero rate, which is removed to reconstruct the signal. Next, high frequency noise is eliminated by first separating the QRS wave groups by the windowing method, determining the number of IMFs at high frequencies using the statistical test method, filtering them out, and reconstructing the signal to obtain a clean signal with the noise removed. The results of this experiment prove that this method is more effective than the original EMD and EEMD methods for removing ECG noise.