综合多目标优化和多准则决策的搅拌过程模拟  

Simulation of mixing process based on integrated multi-objective optimization and multi-criteria decision making

作　　者：武煜坤 李政权 张博群 陈慧敏 王贻得 李凯旋 李明周[1] WU Yukun;LI Zhengquan;ZHANG Boqun;CHEN Huimin;WANG Yide;LI Kaixuan;LI Mingzhou(Jiangxi Key Laboratory of Particle Technology,Jiangxi University of Science and Technology,Ganzhou 341000,China)

机构地区：[1]江西理工大学颗粒技术江西省重点实验室,江西赣州341000

出　　处：《中国粉体技术》2025年第2期15-30,共16页China Powder Science and Technology

基　　金：国家自然科学基金项目,编号:52364047;江西省自然科学基金项目,编号:20212BAB204026。

摘　　要：【目的】基于已开发的计算流体力学(computational fluid dynamics,CFD)-人工神经网络(artificial neural network,ANN)数据预测模型,利用反映不同决策者偏好的多准则决策方法,针对性地解决搅拌釜在不同工业应用中能耗和搅拌效率的均衡需求和特定偏好需求问题。【方法】利用第二代非支配排序遗传算法(non-dominated sorting genetic algorithmⅡ,NSGAⅡ)对CFD-ANN数据预测模型的预测结果进行优化,得到Pareto解集;分别通过熵权法和主观权重对各变量的影响分析确定目标权重占比,并针对不同工业应用场景,利用多准则决策从Pareto解集中选择相应的最优解。【结果】通过优化均衡最优解Opt1,与基础案例Base case相比,能耗降低52.49%,流体混合程度提升1.35%,悬浮均匀性提高72.31%;偏好功率准数Np的最优解降低功耗86.5%,偏好流量准数Nq的最优解达到Pareto解集中的理想状态,偏好σ的最优解将固体浓度标准差降低至Base case的9.93%的同时,也能优化能耗。【结论】基于决策者偏好的多准则决策方法在平衡多个相互冲突的目标方面是有效的。Objective In optimizing the design of stirred tanks,the difficulty lies in the variability of structural parameters,operating condi-tions,and constraints among them.Enhancing performance in one aspect may sacrifice the efficiency of others,making it diffi-cult to achieve systematic optimization and increasing design costs.Striking a balance between maximizing stirring efficiency and minimizing energy consumption is a major challenge in the optimization of stirred tank operation.Based on the developed computational fluid dynamics-artificial neural network(CFD-ANN)data prediction model,a multi-criteria decision-making method that reflects different decision-makers’preferences is used to address the challenge in balancing energy consumption and stirring efficiency of stirred tanks in different industrial applications.Methods The CFD-ANN data prediction model was optimized using non-dominated sorting genetic algorithm II(NSGA II)to obtain the Pareto solution set.The target weight ratio was determined by analyzing the influence of each variable through the entropy weight method and subjective weighting.The corresponding optimal solution was selected from the Pareto solution set for different industrial application scenarios using multi-criteria decision making approach.Results and Discussion Compared with the Base case,the balanced optimal solution(Opt1)reduced energy consumption by 52.49%,increased fluid mixing by 1.35%,and improved suspension uniformity by 72.31%.Decision-makers significantly improved the performance of stirred tanks by adjusting subjective weights,thereby influencing the selection of optimization solu-tions.To ensure industrial standards in key stirred tank parameters,increasing impeller speed and reducing baffle width were recommended.An impeller diameter of 2T/3 with a height of H/4 from the bottom optimized energy-saving,an impeller diameter of T/2.13 with a height of H/6 enhanced uniform fluid mixing,and an impeller diameter of T/1.94 with a height between H/5-H/6 promoted uniform particle su

关 键 词：搅拌釜 计算流体力学 人工神经网络 第二代非支配排序遗传算法 优劣解距离法 

分 类 号：TF301[冶金工程—冶金机械及自动化] TB4[一般工业技术]