文献类型：期刊文献

中文题名：基于多时间尺度协同的大规模原油调度进化算法

英文题名：Multi-timescale cooperative evolutionary algorithm for large-scale crude oil scheduling

作者：张莞婷[1];杜文莉[1];堵威[1]

机构：[1]能源化工过程智能制造教育部重点实验室(华东理工大学),上海200237

年份：2024

卷号：44

期号：5

起止页码：1355

中文期刊名：计算机应用

外文期刊名：journal of Computer Applications

收录：CSTPCD;;北大核心:【北大核心2023】；CSCD:【CSCD_E2023_2024】；

基金：国家重点研发计划项目(2022YFB3305900);国家自然科学基金面上项目(62173144);上海市青年科技启明星计划项目(22QA1402400);上海人工智能实验室资助项目。

语种：中文

中文关键词：进化算法;大规模优化;协同优化;原油调度;多时间尺度

外文关键词：evolutionary algorithm;large-scale optimization;cooperative optimization;crude oil scheduling;multitimescale

摘要：针对原油调度过程存在的资源规模庞大、约束条件复杂、多时间尺度决策衔接困难等问题,提出一种基于多时间尺度协同的进化算法(MTCEA)。首先,根据炼油企业的规模结构和实际需求,建立了一种大规模多时间尺度原油调度优化模型,该模型由面向资源的中长期调度模型和面向操作的短期调度模型构成,通过引入原油资源动态分组策略,实现原油资源的合理配置,以满足不同的调度规模、多时间尺度的特征和精细化生产的要求;其次,为促进不同时间尺度调度决策的融合衔接,设计基于多时间尺度协同的进化算法,并针对不同时间尺度调度模型中的连续决策变量构造子问题进行求解,以实现不同时间尺度调度决策之间的协同优化;最后,在3个实际工业案例进行了算法性能验证。结果表明,与3种具有代表性的大规模进化优化算法(即竞争性粒子群优化算法(CSO)、基于多轨迹搜索的自适应差分进化算法(SaDE-MMTS)和基于混合模型的进化策略(MMES))以及3种高性能混合整数非线性规划(MINLP)数学求解器(即ANTIGONE(Algorithms for coNTinuous/Integer Global Optimization of Nonlinear Equations)、SCIP(Solving Constraint Integer Programs)和SHOT(Supporting Hyperplane Optimization Toolkit))相比,MTCEA的求解最优性指标和稳定性指标分别提高了30%和25%以上。这些显著的性能提升验证了MTCEA在大规模多时间尺度原油调度决策中的实际应用价值和优势。
Aiming to solve the problems of large-scale resources,complex constraints,and difficult cooperation of multitimescale decision-making in the crude oil scheduling process,a Multi-Timescale Cooperation Evolutionary Algorithm(MTCEA)was proposed.Firstly,a large-scale multi-timescale crude oil scheduling optimization model was established according to the scale structure and actual demand of oil refining enterprises,which consists of a resource-oriented mediumand long-term scheduling model and an operation-oriented short-term scheduling model,and achieves a reasonable allocation of crude oil resources through employing a dynamic grouping strategy of crude oil resources to satisfy the requirements of different scheduling scales,multi-timescale characteristics,and fine production.Secondly,to promote the integration of scheduling decisions at different time scales,an evolutionary algorithm based on multi-timescale cooperation was designed and solved by constructing subproblems for the continuous decision variables in scheduling models at different time scales to achieve cooperation optimization between scheduling decisions at different time scales.Finally,MTCEA was verified in three practical industrial cases.Compared with three representative large-scale evolutionary optimization algorithms(i.e.,Competitive Swarm Optimizer(CSO),Self-adaptive Differential Evolution with Modified Multi-Trajectory Search(SaDEMMTS),and Mixture Model-based Evolution Strategy(MMES))and three high-performance Mixed Integer Non-Linear Programming(MINLP)mathematical solvers(ANTIGONE(Algorithms for coNTinuous/Integer Global Optimization of Nonlinear Equations),SCIP(Solving Constraint Integer Programs),and SHOT(Supporting Hyperplane Optimization Toolkit)),the results show that the metrics of the solution optimality and stability of MTCEA are improved by more than 30%and 25%,respectively.These significant performance improvements demonstrate the practical application value and advantages of MTCEA in large-scale multi-timescale crude oil scheduling decisions.