混合多目标粒子群优化算法在热精轧负荷分配优化中的应用
Application of the hybrid multi-objective particle swarm optimization algorithm in load distribution of hot finishing mills

通过对热精轧负荷分配过程的分析, 选取负荷均衡、板形良好和轧制功率最低为目标, 建立了热精轧负荷 分配多目标优化模型. 为了提高多目标优化算法解集的分布性和收敛性, 提出了一种混合多目标粒子群优化算法 (HMOPSO), 该算法根据Pareto支配关系得到Pareto前沿进而保证种群收敛; 采用分解策略维护外部存档, 该策略首 先根据Pareto前沿求出上界点对目标空间进行归一化处理, 然后对种群进行分区处理进而保证种群的分布性能. 仿 真结果表明, HMOPSO的收敛性和分布性都好于MOPSO和dMOPSO; 采用模糊多属性决策的方法从Pareto最优解 集中选择一个Pareto最优解, 通过与经验负荷分配方法相比, 表明该Pareto最优解可以使轧制方案更加合理.
Through the analysis of the process of load distribution of hot finishing mills, a multi-objective optimization model is established with load balancing, good strip shape and minimum power. In order to improve the diversity and convergence performance of Pareto optimal solutions obtained by multi-objective optimization algorithm, a hybrid multi-objective particle swarm optimization algorithm (HMOPSO) is proposed. HMOPSO obtains Pareto front based on the Pareto dominance which can promote population convergence towards Pareto front, and uses the decomposition to maintain external archive by the method of objective space being normalized based on the nadir point of Pareto front and population being partitioned, which can improve the distribution performance of population. Simulation results show that the convergence and distribution performance of the Pareto optimal solutions obtained by HMOPSO are competitive with respect to MOPSO and dMOPSO; the fuzzy multi-attribute decision-making method is adopted to select a Pareto optimal solution from Pareto optimal solution set, and the results show that the solution can get a more reasonable rolling plan compared with the empirical load distribution method.