This paper proposes a new method to optimize cold strip rolling schedule by means of self-adaptive learning based particle swarm optimization (SLPSO). Multiple strategies may be adopted based on their previous behaviors in the searching. This particle swarm optimization version is robust and effective in solving complex problems. Function of power cost was constructed to heuristically direct the SLPSO searching, based on the consideration of power distribution, speed and rolling constraints. The results of simulation demonstrate that SLPSO is more efficient in calculating than others, and provides a new valid method for the intelligent optimum design of scheduling tandem cold strip mill.