In order to further improve the strip’s accuracy and production quality of the hot strip rolling mill, it is necessary to establish appropriate mathematical model and transferring functions of looper system on the basis of coupling process analysis and the practical data. A looper system has the features of nonlinearity, strong coupling, uncertainty, and multiconstraints. Hence, an improved particle swarm optimization algorithm was proposed based on mutation mechanism and particle swarm optimization algorithm in this paper. The PID neural network, trained by improved particle swarm optimization algorithm, was applied to decouple the looper system. The simulation results showed that the algorithm has better control effect than conventional PID control system.