约束多目标粒子群算法在烧结配矿优化中的应用  

Application of constrained multi-objective particle swarm optimization to sinter proportioning optimization

作　　者：冯茜 李擎[1,3] 王耀祖 全威 FENG Qian;LI Qing;WANG Yao-zu;QUAN Wei(School of Automation and Electrical Engineering,University of Science and Technology Beijing,Beijing 100083,China;College of Mechanical Engineering,North China University of Science and Technology,Tangshan Hebei 063210,China;Key Laboratory of Knowledge Automation for Industrial Processes of Ministry of Education,University of Science and Technology Beijing,Beijing 100083,China;Institute of Artificial Intelligence,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]北京科技大学自动化学院,北京100083 [2]华北理工大学机械工程学院,河北唐山063210 [3]北京科技大学工业过程知识自动化教育部重点实验室,北京100083 [4]北京科技大学人工智能研究院,北京100083

出　　处：《控制理论与应用》2022年第5期923-932,共10页Control Theory & Applications

基　　金：国家科技支撑计划项目(2017ZX07402001)资助。

摘　　要：针对钢铁生产的烧结配料过程中铁矿石价格变化大、矿石品位波动、烧结原料信息复杂、烧结配料约束繁多等对烧结配料成本的影响,提出了一种基于区域划分的约束多目标粒子群优化算法对烧结配矿进行优化.以国内某钢铁厂的实际烧结生产配料为例,构建烧结配料系统模型;为了协调全局探索和局部搜索的关系,将自适应角度划分策略融入约束评价准则,结合区域分布提取局部最优解信息,同时引入双外部存储集机制,维护种群多样性;通过标准函数集的测试,验证了所提算法的有效性.该算法应用于配矿过程中,能够兼顾成本与全铁含量,有效降低了烧结配料的成本,对烧结铁矿石资源的综合利用及质量保证具有重要意义.Aiming at the influence of iron ore price change,ore grade fluctuation,complex information of sintering raw materials and numerous constraints of sintering batching in iron and steel production,a constrained multi-objective particle swarm optimization algorithm based on region division is proposed to optimize the actual sintering production burden of a domestic iron and steel plant.The sintering burden system model is established.In order to coordinate the relationship between global development and local search,the adaptive angle division strategy is integrated into the constraint evaluation criterion,the local optimal solution information is extracted combined with regional distribution,and the two-archive mechanism is introduced to maintain population diversity.The effectiveness of the proposed algorithm is verified by the test of standard function set.The algorithm is applied to account the cost and total iron content in the process of ore blending,effectively reducing the cost of sintering ingredients.It has important significance for the comprehensive utilization of sintered iron ore resources and quality assurance.

关 键 词：烧结配料 多目标粒子群算法 约束优化 区域划分 

分 类 号：TF046.4[冶金工程—冶金物理化学] TP18[自动化与计算机技术—控制理论与控制工程]