Study On Time Series Analysis Prediction Method Based On Wavelet Analysis

Wavelet transform is a localization analysis of the time and frequency of the signal.In addition,it can decompose the signal at high frequency and low frequency,So as to better analyze and process the signal.Regression analysis is one of the most extensive methods to study the relationship between independent variables and dependent variables,which can accurately reflect the relationship between data.Combining the above two data analysis methods can establish a wavelet analysis-regression model to perform timefrequency analysis on the data,and we can also make further use of the time series method for prediction.With the development of the big data era,data processing has received more and more attention.Time series prediction method is one of the most effective methods to deal with time-varying data.In the process of time series prediction,wavelet can effectively extract important information,and the signal or series can be analyzed in multi-scale by stretching and translation operations,and then focus on any details of the analyzed object.therefore,it is more suitable to deal with non-stationary time series,and better prediction results can be obtained.In recent years,more and more people begin to use wavelet to analyze and predict time series,find combination methods and models,and gradually improve the theoretical basis.This paper makes use of the above advantages to establish a combination model for analysis and prediction.the main research contents are as follows:(1)Based on the multi-resolution analysis theory of wavelet analysis,a wavelet analysis-regression model is established.first,the appropriate wavelet function is selected to decompose the signal or data,then the decomposed wavelet coefficients are reconstructed,and finally the regression relationship between the signal and independent variables is obtained.(2)Based on the wavelet decomposition and reconstruction algorithm,the wavelet analysis-time series model is established.firstly,the abnormal value of the signal or data is detected,then the wavelet decomposition and reconstruction algorithm is used to denoise the data,and the appropriate time series model is selected for analysis and prediction.finally,the appropriate test method is selected to test the residuals,parameters and models,and the confidence of the prediction results is calculated.(3)The above model is used for example analysis In this paper,taking the coloring,fading and wavelength data of WO₃ at a certain PH value as an example,the wavelet analysis-regression model is used to analyze the transmittance of WO₃ nano-block thin films,and the variation law with wavelength is obtained.Then the wavelet analysis-time series model and the wavelength change law are used to predict the transmittance.After examination,the prediction results are feasible and effective.In addition,the global epidemic situation of COVID-19 is grim.Taking the daily data of new pneumonia cases in eastern and western countries as examples,this paper uses ARIMA model and wavelet analysis-time series model to predict the future epidemic trend of the two countries respectively.The test results of the two prediction models show that the wavelet analysis-time series model is more accurate and reliable than the ARIMA model,and the result is feasible and effective.