Abstract: To efficiently solve the scheduling problem of material delivery for automotive assembly lines based on milk-run delivery， a scheduling method was developed by the modified immune clone selection algorithm. Firstly， a mathematical programming model was set up with an objective function of minimizing total inventory for all stations over the planning horizon. Then， a modified immune clone selection algorithm was developed to solve the proposed problem. Both the simulated annealing operator and neighborhood search operator were applied to clone population and memory vault， respectively， in the design of algorithm. It overcomes deficiencies of the traditional immune clone selection algorithm， such as tendencies to trap into local optima and limited search depth. Finally， the simulation experiments were carried out and the results indicate that the as-proposed algorithm is valid and feasible.

Key words: milk-run delivery, scheduling, supermarket, immune clone selection algorithm, neighborhood search