Research Of ECG Signal Prediction Method Of Neural Network Based On Phase Space Reconstruction

Cardiovascular disease is a relatively high incidence of disease in modern society,when a patient needs to monitor ECG,the amount of data needs to be collected.The data needs to be stored,so a large amount of data will take up a lot of space.Therefore,requirement of the storage is relatively high,in the process of data transmission will waste a lot of energy.For the situation under the line,the hospital needs to archive each patient's data,so large storage space is needed.If the data can be predicted,the hospital will be able to store some of the data,other data can be used to predict these values,but also saves the storage space.In the process of signal acquisition,it can be predicted and does not need to collect data all the time,which saves the energy of the collection and transmission of data.Prediction is one of the most widely used technology in data mining,it is a speculation of objective phenomenon.Artificial neural network(ANN)is a widely used tool for forecasting data,which has strong learning ability and adaptive ability.BP(Back Propagation)neural network is a kind of artificial neural network,it is suitable for dealing with the nonlinear system of unknown background knowledge and unclear rule.ECG signal is a non-stationary and nonlinear signal.In view of the fact that input nodes of BP neural network are uncertain,based on the prediction of ECG signal.In this paper the combination of phase space reconstruction and BP neural network to predict ECG signal.Calculation of the time delay and embedding dimension of the reconstructed phase space with mutual information method and Cao's method,two parameters are used to change the input data of the network of the phase space.The data of reconstitution is input to the network for training until the predictive value is goal.In this paper,MATLAB is used as the tool of simulation,the ECG signal provided by arrhythmia database of Massachusetts Institute of Technology is used as the simulation signal.The result shows that the model can predict the signal in the range of error tolerance.The algorithm objectively determines the number of nodes in the network and overcomes the influence of subjective factor on the prediction result.It can be used as a reference for the prediction of ECG signal and it is great significance to the prevention of disease in the future.