炼油厂原油处理短期生产计划调度优化
本文关键词: 短期生产计划 原油处理 启发式 多目标优化 遗传算法 出处:《广东工业大学》2016年博士论文 论文类型:学位论文
【摘要】:炼油工业是国民经济发展的重要支柱产业。炼油生产计划包括长期生产计划和短期生产计划。长期生产计划属于战略性计划,由于可以看成连续参数系统优化问题,可以用数学规划的方法解决,主要是利用线性规划进行建模和求解。目前对长期生产计划优化的理论研究已有成熟的方法。炼油短期生产计划和调度问题,不仅要处理离散事件过程,还要处理连续变量,为了使炼油厂高效运转,不仅需要优化离散事件执行的顺序,同时还要对连续变量的值进行优化。但是,在炼油生产过程的调度中,事先并不知道需要调度的具体事件,这些生产作业需要在计划过程中产生。因此,其复杂性和困难程度远远大于对离散过程和批处理过程的调度。本课题组在前期研究中,证明了炼油短期生产计划问题属于NP-hard问题,这就排除了用精确方法求解该问题的可能性。另外,在获得炼油短期生产计划的过程中,人们不仅需要定义计划周期内的作业,同时还需要对这些作业进行排序,所以不能直接用启发式和智能优化方法来解决该问题。原油处理短期生产计划问题是炼油短期生产计划中最困难的问题之一,因此本文仅研究原油处理短期生产计划问题。由于启发式方法和数学规划方法都不能直接应用于炼油短期生产计划问题,本课题组另辟蹊径,从控制理论角度,将原油处理短期生产计划问题分解为上下两层:在上层求炼油生产计划以优化相关目标,在下层则求解一个详细的短期计划以实现上层炼油计划。在课题组的前期研究中,已经成功地用基于线性规划的方法解决了上层炼油计划的优化求解问题,但是下层详细生产计划优化亟待解决。因此,本文研究在上层炼油计划已知的情况下,研究下层详细生产计划优化问题,主要进行了以下几方面的研究工作:(1)由于数学规划和启发式方法都不能直接地应用于原油处理短期生产计划问题,因此需要另找一种方法来对问题进行求解,这是本文研究的初衷。首先建立了所研究问题的数学规划模型,基于此模型分析问题的本质特点。定义了原油处理短期生产计划问题是由一系列的运作决策构成,从控制理论角度,将原油处理详细生产计划问题转换为供油罐到蒸馏塔的指派问题。这样一来,使得启发式算法和智能算法可以用于解决原油处理短期生产计划问题,从而克服用数学规划方法的计算复杂性难题。(2)提出了两种启发式算法对原油处理详细生产计划问题进行求解,两种方法均能保证给定的上层蒸馏塔炼油计划一定能实现。虽然启发式算法简单,但不能保证解的最优性,不过实例结果表明,本文所提出的启发式算法对原油处理详细生产计划优化问题还是有效的。(3)原油处理详细生产计划问题涉及到多个优化目标,本文通过加权和将多目标优化问题转换为单目标优化问题,提出了适合问题求解的单目标遗传算法。基于可调度性条件,为了确保解的可行性,提出了一个转换算法确保所有随机产生的染色体都能对应一个可行的详细生产计划。并用实例验证了算法的有效性。(4)提出了解决原油处理详细生产计划的多目标优化算法。原油处理详细生产计划阶段涉及到的多个目标是相互冲突的,本文求解基于帕累托最优的多目标优化解。设计了一个长度可变的双基因染色体来描述下层详细炼油计划,经过巧妙的解码,使得任何一个随机产生的染色体不需要转换正好对应一个可行的详细生产计划。接着,基于改进的非支配排序遗传算法对问题进行求解,并用实例验证了算法的有效性。本文将原本不能用启发式和智能算法求解的原油处理短期生产计划问题转换为供油罐到蒸馏塔的指派问题,使得启发式算法和智能算法可以用于解决原油处理短期生产计划问题,从而克服用数学规划方法的计算复杂性,为开发实际可用的优化方法提供了途径,解决了其中的一个重要难题。
[Abstract]:Oil refining industry is an important pillar industry of the national economy. Petroleum production plan including long-term production planning and short-term scheduling. Long term production plan is a strategic plan, as can be seen as a continuous parameter optimization problem can be solved by the system, the method of mathematical programming, mainly using linear programming modeling and solving. The long-term production planning optimization theoretical research has been mature. Refinery production planning and scheduling problem in the short term, not only to deal with the discrete event process, to deal with continuous variables, in order to make the refinery efficient operation, not only need to optimize the order of execution of discrete events, but also on the values of the continuous variables were optimized. However, in the process of oil refining production scheduling in advance do not know the specific needs of event scheduling, these operations need to produce in the planning process. Therefore, its complexity and difficulties Difficult is far greater than the discrete process and batch process scheduling. This group in the previous study, proved that the refinery short-term scheduling problem belongs to NP-hard, which precludes the possibility of solving the problem with exact methods. In addition, in the short term production plan for oil refining process, people not only need to define the plan within the period of work, but also need to sort these jobs, so you cannot directly use the heuristic and intelligent optimization method to solve the problem. The short-term crude oil processing production planning problem is one of the most difficult problems in the production of refinery short-term plan, so we only study the short-term crude oil processing production planning problem. The heuristic method and mathematical programming method can not be directly applied in refinery short-term scheduling problem, the research group and from the perspective of control theory, the crude oil processing of short-term production Planning problem is decomposed into two layers: in the upper for petroleum production plan to optimize the target, in the lower layer for a detailed plan to achieve short-term upper refining plans. In the early research group, have successfully used the method based on the linear programming optimization problem has been solved on the layer of oil refining project. But with lower production plan optimization to be solved. Therefore, this paper studies on the upper refining schedule is known, optimization of production plan with lower research, the main research work are as follows: (1) by mathematical programming and heuristic methods cannot be directly applied to crude oil processing short-term production planning problem, therefore need to find a way to solve the problem, this is the original intention of this paper. Firstly, mathematical programming model of this model is established, based on analysis of Q The essential characteristics of the subject. The definition of crude oil processing short term production plan is composed of a series of operational decisions, from the perspective of control theory, the crude oil processing detailed production planning problem into a fuel tank to the assignment problem of distillation tower. As a result, the heuristic algorithm and intelligent algorithm can be used to solve the problem of short-term crude oil processing production plan thus, the computational complexity of the problem of taking g mathematical programming method. (2) proposed two heuristic algorithm to solve the production planning problem with crude oil processing, the two methods can ensure the upper crude oil distillation column is given plan will be achieved. Although the heuristic algorithm is simple, but can not guarantee the optimality of the solutions, but the results show that the heuristic algorithm proposed in this paper for crude oil processing detailed production planning optimization problem is effective. (3) crude oil processing with production plan problem Involves multiple optimization objectives through weighted and multi-objective optimization problem is transformed into a single objective optimization problem, proposed a single objective genetic algorithm for solving the problem. Based on the schedulability conditions, in order to ensure the feasibility of the solution, proposed a conversion algorithm to ensure that a randomly generated chromosome can correspond to a with feasible production plan. And verify the validity of the algorithm by an example. (4) is proposed to solve the multi-objective optimization algorithm with crude oil processing production plan. The crude oil processing multiple targets with the production planning stage is related to the conflict, this paper is based on Pareto optimal solutions of multi-objective optimization design of double. A variable length chromosome to describe with lower refining plans, through clever decoding, so that any a randomly generated chromosome without conversion corresponds to a With the production plan. Then, non dominated sorting genetic algorithm to solve the problem based on the improved and verified the validity of the algorithm with examples. This paper originally cannot use heuristic and intelligent algorithms for solving the problem of short-term crude oil processing production plan into fuel tank to the distillation tower assignment problem, the heuristic algorithm and intelligent the algorithm can be used to solve the problem of short-term crude oil processing production plan, and the computational complexity of mathematical programming method taking g, provide a way for the optimization method of practical development, solve one important problem.
【学位授予单位】:广东工业大学
【学位级别】:博士
【学位授予年份】:2016
【分类号】:F273;F426.72
【相似文献】
相关期刊论文 前7条
1 樊昶明;伍乃骐;;原油处理过程短期生产计划的辅助仿真[J];工业工程;2013年02期
2 支淑民;陈刚田;万世清;王艳;;坨四站原油处理的影响因素及对策研究[J];价值工程;2013年18期
3 白丽平;伍乃骐;;炼油厂原油处理过程短期生产计划及其复杂性[J];工业工程;2011年01期
4 吴月先;新疆陆梁油田百万吨级原油处理站正式破土动工[J];新疆石油地质;2001年06期
5 冯耀忠;简介俄罗斯降低集油和原油处理费用的途径[J];油田地面工程;1997年05期
6 丁成杰;伍乃骐;白丽平;;基于Petri网原油处理短期生产计划可调度性分析[J];工业工程;2012年06期
7 ;[J];;年期
相关会议论文 前2条
1 崔营双;王伟华;郭昌文;;系统工程原理在桩西联原油处理中的应用[A];2005年山东省石油学会营销专业委员会论文选编[C];2005年
2 邓燕;许建华;;王庄油田稠油高效低耗处理工艺研究[A];山东石油学会稠油特稠油地面集输与处理技术研讨会论文集[C];2009年
相关博士学位论文 前1条
1 侯艳;炼油厂原油处理短期生产计划调度优化[D];广东工业大学;2016年
相关硕士学位论文 前4条
1 王新亮;海洋石油原油处理设备及其电仪加工设计[D];西南石油大学;2014年
2 耿亚;基于JIT的炼油生产原油处理短期生产计划方法与实现[D];广东工业大学;2006年
3 黎明;炼化企业原油处理过程短期生产计划的两阶段求解方法[D];广东工业大学;2012年
4 张思维;混合供油模式下炼油企业原油处理过程的Petri网建模和短期生产计划分析[D];广东工业大学;2015年
,本文编号:1486167
本文链接:https://www.wllwen.com/shoufeilunwen/jjglss/1486167.html