Abstract: This paper proposes a new method of fast modeling for computing the residual sum of squares, determinating the model order and estimating the parameters by traditional least squares.
The residual sum of squares can bo easily computed before parameter estimation. Thus, the computing work for establishing a model with unknown order by this method is nearly equal to that with known order. Alsc this method requires lass storage capacity., so that the modeling work can be performed even with a pocket microcomputer(e. g. FX-702P).