The restraint combination optimization question of the cold rolling production scheduling is typical NP-hard problem, which included the width, thickness, degree of hardness jump, product stock and tows the penalty of the rolling tube blank. To establish minimum value model of the cold rolling production batch scheduling problem based on TSP, and then proposed Improved Ant Group Algorithm which combined ant group optimization algorithm with the immunity algorithm (AIA) fusion algorithm to deal with it. Applying in a flexible way which inoculated the most superior ant of each generation against the vaccine (prior-knowledge), algorithm has been weal dealt with restraints. In view of result in the industry data simulation: the solution based on improved ant group immunity algorithm is obviously better than the solution base on the sole improvement ant group optimization algorithm, the former should be weal used and spread.