基于遗传算法的铸造热处理炉次计划  

Casting heat treatment charge plan based on genetic algorithm

作　　者：叶虎[1] 计效园[1] 周建新[1] 陈飞[1] 

机构地区：[1]华中科技大学材料成形与模具技术国家重点实验室,湖北武汉430074

出　　处：《计算机集成制造系统》2017年第3期575-583,共9页Computer Integrated Manufacturing Systems

基　　金：国家数控重大专项资助项目(2012ZX04012-011)~~

摘　　要：为了提高铸造热处理生产效率,从合炉约束、炉次容量利用率以及交货期3个方面综合考虑,建立了铸造热处理炉次计划多目标整数规划模型,并提出了分类与遗传算法相结合的求解方案。基于合炉约束将任务集分类并生成炉次计划可能的候选集,根据炉次容量利用率及交货期两个因素对候选集评分,将排名前5的候选集作为炉次计划的最终候选集。设计了改进的遗传算法,改进策略包括采用两种不同交叉算子,模拟退火机制以及重置算子。基于改进的遗传算法对5个候选集进行求解,将得到的最优方案作为最终炉次计划。通过仿真实验与实际对比验证了数学模型及求解算法的有效性和适用性。To improve the efficiency of casting heat treatment,a multi-objective integer programming model was presented by taking furnace combining constraints,furnace capacity utilization and delivery deadline into account,and the solving method by combining genetic algorithm with classification was also proposed.Based on furnace combining constraints,the proper candidate sets of charge plan were generated with classification.Each candidate set was evaluated by utilization of furnace and delivery,and the top five of candidate sets were selected as the final candidate sets for charge plan.The genetic algorithm was improved by adopting two different crossover operators,simulated annealing mechanism and reset operator.The optimal solutions for five candidate sets were obtained by the improved genetic algorithm,and the best of which was considered as the optimal charge plan.The effectiveness and applicability of the proposed model and solving algorithm were demonstrated by simulation and comparison with production practice.

关 键 词：热处理 炉次计划 分类 遗传算法 模拟退火 

分 类 号：TP39[自动化与计算机技术—计算机应用技术] TG28[自动化与计算机技术—计算机科学与技术]