An ECG signal analysis and prediction method combined with VMD and neural network

Abstract

A new data prediction method for electrocardiogram (ECG) signals is proposed in this paper, which combines error back propagation neural network (BPNN) and variational mode decomposition (VMD) technology. The proposed method involves three parts. First, with VMD applied, the ECG signal which contains baseline wander (BW) noise is decomposed into a set of modes. Second, by analyzing the center frequency of each mode, the modes are divided into feature modes and noise modes which corresponding ECG signal and BW noise respectively. Third, the feature modes are used as the inputs of BPNN for data prediction. By learning and training the network, the weights and thresholds are identified, finally achieved the purpose of ECG signal prediction. Simulation results show that the proposed method is effective and has good performance in several performance indicators.