Time Series Prediction Method Based On Optimization Theory And Wavelet Analysis And Its Application

As one of the branches of Probability Statistics,time series analysis is analyzed and dealt with the ordered random dynamic data observed value.In other words,it is selected the suitable mathematical model to simulate the time series and forecasted its trend by the model we built.In the practical application,time series often present in highly sophisticated nonlinear states.The traditional forecasting methods can not meet the requirements of higher prediction accuracy,so an improved parameter estimation optimization algorithm and a new forecasting method combining wavelet denoising and neural network model are proposed in this paper,and the efficiency of methods are proved by examples.The wavelet denoising is used to preprocess the history time series data of the stock.Then,the preprocessed series are modeled and predicted by neural network models.Firstly,the paper introduces the related theories of time series analysis,optimization method,wavelet analysis and neural network,and analyses the conjugate gradient method in detail,then summarizes the development history and research status of wavelet analysis and neural network.Secondly,the paper establishes an improved conjugate gradient algorithm and successfully applied to the time series model parameter optimization,namely the use of an improved conjugate gradient method to optimize the parameters of the model.The numerical results prove the effectiveness of the algorithm,and the time series example shows the superiority of the method.Thirdly,based on the theory of wavelet threshold denoising,a new threshold function and threshold formula are proposed,then the paper put forward a new wavelet threshold denoising algorithm.And the simulation results for noise signals show that the new method improves the signal-to-noise ratio and reduces the ]root-mean-square error,which prove the feasibility of the algorithm.Finally,on the basis of the new wavelet denoising algorithm,by combining with the neural network model,a new combination forecasting method is proposed.At first,the de-noising pretreatment is carried out for the Shanghai composite index closing price.And then we apply the new method to model and predict the series,the result shows that the new prediction method has better prediction effect.