请问直接使用matlab里面的工具箱工具建立AR，ARMA模型行不？

jzc1983

谢谢

eight

原帖由 jzc1983 于 2007-4-7 10:01 发表
谢谢

请搜索版面，学会提问前先动手

笑石头

AR   Computes AR-models of signals using various approaches.
    Model = AR(Y,N)  or  TH = AR(Y,N,Approach)  or TH = AR(Y,N,Approach,Win)

    Model: returned as an IDPOLY model with the estimated parameters of the
           AR-model, see HELP IDPOLY.

    Y: The time series to be modelled, an IDDATA object. (See HELP IDDATA)
    N: The order of the AR-model
    Approach: The method used, one of the following ones:
       'fb' : The forward-backward approach (default)
       'ls' : The Least Squares method
       'yw' : The Yule-Walker method
       'burg': Burg's method
       'gl' : A geometric lattice method
       For the two latter ones, reflection coefficients and loss functions
       are returned in REFL by [Model,REFL] = AR(y,n,approach)
       If any of these arguments end with a zero (like 'burg0'), the
       computation of the covariance is suppressed.
    Win : Windows employed, one of the following ones:
       'now' : No windowing (default, except when approach='yw')
       'prw' : Prewindowing
       'pow' : Postwindowing
       'ppw' : pre- and post-windowing


    The Property/Value pairs 'MaxSize'/maxsize and 'Ts'/Ts can be added to
    set the MaxSize property (see also IDPROPS ALG) and to override the sampling
    interval of the data: Example: Model = AR(Y,N,Approach,'MaxSize',500).

    See also IVAR and for the multi-output case ARX and N4SID.

Overloaded methods
    help xregcovariance/ar.m