Nonlinear Signal Prediction Based On Empirical Mode Decomposition And Neural Network

The prediction of nonlinear signals has always been an important branch of signal processing.With the in-depth study of nature,more and more signals are considered as nonlinear signals,such as underwater acoustic signals,sunspot numbers,and speech signals.Establishing a predictive model for nonlinear signals can better reveal the nature and laws of things and have important research significance and practical value for guiding human life.Empirical mode decomposition(EMD)has been developed in recent years as a new data analysis method.In this paper,the original signal is decomposed by using EMD and its improved method,and the wavelet neural network and extreme learning machine neural network are used to predict.Find the advantages of the combined forecasting model in nonlinear signal prediction,and the sunspots and the underwater acoustic signals will be predicted by prediction models.The main contents and innovations are as follows:(1)The combined prediction model and the wavelet neural network are used to establish the combined forecasting model.By comparing the three empirical mode decomposition methods with the wavelet neural network,the monthly average value of sunspots is predicted.The experimental results show that the complete empirical mode decomposition can effectively avoid the modal aliasing phenomenon.Combined with the wavelet neural network,it can effectively predict the sunspot number.It is a better algorithm.(2)A combined forecasting model based on wavelet neural network optimized by fruit fly optimization algorithm combined with complete empirical mode decomposition is proposed,and forecasting the average monthly value of sunspots.The fruit fly optimization algorithm can effectively find out the weights in the wavelet neural network neurons and effectively improve the prediction accuracy.The experimental results show that compared with the neural network which is not optimized,the prediction model has a certain improvement in the prediction accuracy.(3)A hybrid prediction model based on extreme-point symmetric mode decomposition and extreme learning machine is proposed,and the underwater acoustic signal is predicted.Extreme learning machine training speed,setting parameters is simple.Experimental results show that the model can greatly improve the prediction accuracy.