Based　on　the　clonal　selection　theory,　the　main　mechanisms　of　clone　in　immune　system,　which　will　be　explored　in　the　field　of　artificial　intelligence,　are　analyzed　in　this　paper.　A　new　artificial　immune　system　algorithm,　Antibody　Clone　Algorithm,　is　put　forward.　Compared　with　an　improved　gene　algorithm,　the　new　algorithm　is　shown　to　be　an　evolutionary　strategy　capable　of　solving　complex　machine　learning　tasks,　like　multi-modal　optimization.　And　using　the　theories　of　Markov　Chain,　it　is　proofed　that　the　antibody　clone　algorithm　is　convergent.