基于偏好的多目标粒子群优化的结晶器预测控制

针对结晶器出口温度和液位控制问题，提出了一种基于改进的偏好多目标粒子群优化的非线性预测控制算法（IMPSO-NPC）。改进的偏好多目标粒子群优化算法（IP-MPSO）将参考点偏好算法和参考区域偏好算法融合在一起，在参考点和参考区移动过程中动态调整参考区，控制解集的偏好范围。另外，为了选取粒子群全局最优粒子，提出一种球扇占优的策略，提高了粒子群的搜索能力。将改进算法应用于结晶器的控制过程，仿真结果证明了其有效性和可行性。

Mold predictive control based on preference multi-objective particle swarm optimization

A nonlinear predictive control algorithm based on a preference multi-objective particle swarm optimization algorithm (IMPSO-NPC) has been proposed in an attempt to control problems of mold level and temperature at the mold. An improved preference multi-objective particle swarm optimization (IP-MPSO) algorithm combines a reference region with a reference point in order to guarantee the preference direction. In the process of moving reference regions and reference points, IMPSO-NPC dynamically adjusts the reference regions and controls the preference range. In order to further improve the search performance, spherical sector dominance is proposed for gBest of PSO. The proposed method can be applied to mold control systems and simulation results show that the new method is feasible and effective.