摘要
针对热轧生产过程中工艺参数优化问题,提出了一种差分进化算法。首先,构建随机森林回归模型预测质量特征;然后,基于随机森林回归模型的工艺参数重要性排序结果选择待优化的工艺参数;最后,利用差分进化算法优化工艺参数并使用最优结果替换待优化的工艺参数。文章提出以缺陷相对发生率平均降低值和平均降低率作为评价指标,对比研究优化前后的结果,与其他优化算法相比,差分进化算法达到了最优性能。应用分析表明,所提算法进行工艺参数优化后,单位板坯中的缺陷个数均少于25,算法执行平均用时为8.57 s,具有良好的可操作性与应用推广价值。
In order to address the optimization of parameters in hot rolling production processes,a differential evolution algorithm is proposed.Firstly,a random forest regression model is constructed to predict quality characteristic values.Then,based on the feature importance ranking results of the random forest regression model,the process parameters to be optimized are selected.Finally,the differential evolution algorithm is used to optimize the parameters and the optimal results are used to replace the process parameters to be optimized.The average reduction value and average reduction rate of defect occurrence rate are proposed as evaluation indicators.Comparative studies of the results before and after optimization show that compared to other optimization algorithms,the differential evolution algorithm achieves optimal performance.Application analysis demonstrates that after optimizing process parameters with the proposed algorithm,the number of defects in the unit slab is less than 25,and the average execution time of the algorithm is 8.57 s,indicating its good operability and practical value.
作者
王昊
包向军
周剑波
汪晶
张超
Wang Hao;Bao Xiangjun;Zhou Jianbo;Wang Jing;Zhang Chao(Dalian University of Technology;Anhui University of Technology;Shougang Changzhi Iron and Steel Group Co.,Ltd.;Shanghai Baosight Software Co.,Ltd.)
出处
《冶金能源》
北大核心
2024年第4期48-53,共6页
Energy For Metallurgical Industry
基金
国家重点研发计划资助项目(2020YFB1711104)。
关键词
热轧
工艺参数优化
差分进化
随机森林
hot rolling
process parameter optimization
differential evolution
random forest
