A Class Of Bilinear Time Series Model Parameter Estimation And Test

Bilinear time series model is a kind of nonlinear time series, because of the presence of bilinear terms, which makes the research of problem of great complexity. In the model, parameter estimation and test is the important research content. In this paper, a class of bilinear time series model is researched, using different methods on double-track line model parameter estimation and test.This paper focuses on the discussion of bilinear time series model BL (2,0,1,1).It describes its stable condition and reversible condition, and use the traditional linear prediction method to study the model BL (2,0,1,1) of the conditional expectation of prediction,getting the prediction results. On the assumption that under stationary conditions, respectively the moment estimation method, the MCMC estimation method, and used to model ARMA linear estimation of the maximum likelihood estimation method are being on model BL (2,0,1,1) of the parameter estimation. And the maximum likelihood estimate of the simulation, and the simulation result is satisfactory. The method is illustrated with a high estimation precision estimation.The graph model is applied to the bilinear time series model BL (2,0,1,1) parametric test, the model is expressed as a time series of the chain graph model. Application of knowledge to the parameter of the test, and a randomly generated sample data, in the given parameters, the nonlinear coefficient test, the simulated result is good, and the promotion of use of bilinear time series model BL (p,0, r, s). Solution to the model is expressed as a graph problem----dealing with nonlinear term...