earticle

영어

This paper uses phase space reconstruction as the basis of the multi – input nonlinear method. It is obvious that for a system with multiple variables, it is necessary to choose the involved variable before reconstruction, and identify the reconstruction parameter, after determining the reconstructed variable, in order to complete the basic reconstruction. Therefore, based on the selection of the nonlinear correlation, this paper introduces the method for choosing the correct input variable at first , and then introduces some commonly used methods to identify the reconstruction parameter, such as mutual information method, auto – correlation method and average displacement method etc.. Furthermore, it specially introduces the C – C method. By carrying out the multivariate combination forecasting simulation for time series of the Lorentz Equation and comparing the reconstruction phase diagram of the multivariate phase space, this paper verifies the accuracy of selecting reconstructed input vector based on nonlinear correlation and the effectiveness of using C- C method to decide the reconstruction parameter.