Research Of ECG Signal Prediction Method Of Neural Network Based On Variational Mode Decomposition

With the rapid development of modern society,network technology and database technology have made great progress,people need to get information quickly and concisely,and they need scientific forecasting to help and guide their behavior.People are eager to predict some physical disease precisely such as angiocardiopathy.Thereout a variety of prediction algorithms and tools are derived.Artificial neural networks is one of the best.ANNs is a nonlinear model to simulate the processing function of biological brain,it has a great deal of excellent characteristics just like plasticity,robustness and ability of information processing.The error back propagation neural network is one of the most typical branches of ANNs,it is especially suitable for the prediction of nonlinear and non-stationary signals.Electrocardiogram signal is such a signal.It can clearly reflect the activities of heart,making doctor understand the condition fleetly,so that they can treat patients quickly and accurately.The forecast of ECG signal is significant.In this paper,BPNN is used to predict the ECG signal.BPNN also has some defects,in order to improve the predictive value and obtain a better prediction result,we use a new signal processing technology-variational mode decomposition,to improve the input of BPNN in the prediction process.This method objectively determined the number of input nodes,avoid the influence of subjective factors on it,so that make BPNN be improved.Variational mode decomposition is a new,entirely non-recursive signal decomposition method,it can decompose the given signal into a set of modes which around the center frequencies.In this paper,we decompose the input ECG signal which contains the baseline shift noise into a set of modes with VMD,by analyzing the center frequency of each mode,we complete the selection of feature mode and noise removal.Then the rest of the mode is used as the input signal of the BPNN.By learning and training network,the weight and threshold is identified,finally achieved the purpose of data prediction.In this paper,we use MATLAB as a simulation tool,use the ECG signals of No.100 th and No.119 th as simulation signals which are provided by arrhythmia database of Massachusetts Institute of Technology.Simulation experiments show that the improved algorithm is effective and has good performance in the case of error tolerance,the validity of ECG signal prediction of the improved algorithm is verified.