Method of moment (MoM) expends large computer resource when calculating radiation and scattering problem of large target. The adaptive cross approximation (ACA) algorithm is applied to compress the distant field rank-deficient sub-matrices of the impedance matrix through linear algebra manipulations, which reduces memory needs and calculation complexity in MoM when computing electrically large problem. The RWG (Rao-Wilton-Glisson) triangle spaces are grouped according to the Non-Uniform Rational B-Splines (NURBS) spaces by means of NURBS modeling method. The grouping method is designed to realize the ACA algorithm based on MOM, which could make the subgroups more regular and clearer on the boundary line. Numerical examples are performed, the results show that this method can be the same with MoM in computation precision, and simultaneously the memory and compute time are reduced by this method. RCS (Radar Cross Section) of electrically large metal missile calculated by using this method is in good agreement with the result calculated by the commercial software, which shows that the algorithm is accurate and effective.