Abstract: In traditional immune algorithm slow convergence speed,parameter setting complex and affinity calculation error existing shortcomings,using antibody detection radius R_probe detect antibody concentration strategy,the article puts forward an improved immune algorithm,takes antigen and antibody respectively corresponding to the solution space X and a solution x_i,X be divided into K_opt cluster,each cluster head with a radius R_probe of indrawn all messages to perceive its surrounding antibody,the antibody concentration calculation method can achieve real nonsymmetrical mutation and can improve the algorithm convergence speed and has the escaping ability from local minimum.