多目标优化算法及其在化工中的应用研究

发布时间：2018-12-29 17:58

【摘要】：多目标优化算法广泛应用于化工领域,诸如过程控制与操作优化、化工设备设计、环境工程等。近年来,越来越多的学者将多目标优化算法与流程模拟器相结合解化工优化问题。由于流程模拟需要花费大量时间进行收敛计算,因此,优化算法必须能够在较少的目标函数评价次数的情况下,也能快速达到收敛。非支配遗传算法是多目标优化领域应用最广的算法,但其必须经过上万次的目标函数评价才能得到较好的结果,而且其本身也存在容易过早收敛、局部搜索能力不强等缺点。因此,本文旨在提出一种高效的多目标优化算法,将其应用于化工过程的优化。本文的工作主要有以下几个方面:(1)阐述了多目标优化算法的研究背景和意义,从科学研究和工程应用两个方面,介绍了多目标进化算法的发展,并对列队竞争算法的研究和应用进展进行了简单介绍;(2)介绍了多目标优化问题的相关概念与定义,详细描述了列队竞争算法LCA和非支配遗传算法NSGA-II的计算流程与关键算子,引入了解集收敛性和均匀性评价指标;(3)提出了一种多目标列队竞争算法MOLCA,采用多种策略,在降低目标函数评价次数的同时达到快速收敛。对MOLCA主要参数的设定进行了讨论,然后利用经典测试函数对MOLCA进行测试分析,与NSGA-II相比,该方法表现更优秀。将MOLCA应用于催化裂化主分馏塔的操作参数优化,以总经济效益和系统能耗为两目标,给出了优化的操作方案;(4)针对NSGA-II存在容易收敛于局部最优解和计算时间久的问题,提出了一种基于多目标列队竞争算法和非支配遗传算法的混合算法MOLCA-NSGA-II。经过经典测试函数的测试分析表明,该算法在运算时间、收敛性和分布性方面都要明显优于NSGA-II。将MOLCA-NSGA-II应用于甲醇制烯烃分离过程优化,结果给出了一系列Pareto最优解,可以根据不同的生产要求,综合考虑能耗和收率,选择适宜的操作条件。
[Abstract]:Multi-objective optimization algorithm is widely used in chemical engineering fields, such as process control and operation optimization, chemical equipment design, environmental engineering and so on. In recent years, more and more scholars combine multi-objective optimization algorithm with process simulator to solve chemical optimization problem. Because the process simulation takes a lot of time to calculate convergence, the optimization algorithm must be able to achieve convergence quickly with less evaluation times of objective function. The non-dominated genetic algorithm is the most widely used algorithm in the field of multi-objective optimization, but it must pass through tens of thousands of objective function evaluation to get a better result, and its own shortcomings such as easy premature convergence, weak local search ability and so on. Therefore, this paper proposes an efficient multi-objective optimization algorithm and applies it to the optimization of chemical processes. The main work of this paper is as follows: (1) the research background and significance of multi-objective optimization algorithm are expounded, and the development of multi-objective evolutionary algorithm is introduced from two aspects: scientific research and engineering application. The research and application of queue competition algorithm are briefly introduced. (2) the concept and definition of multi-objective optimization problem are introduced, the calculation flow and key operators of LCA and NSGA-II are described in detail, and the evaluation index of convergence and uniformity of solution set is introduced. (3) A multi-objective queue competition algorithm (MOLCA,) is proposed, which adopts many strategies to reduce the number of evaluation of the objective function and achieve rapid convergence. The setting of the main parameters of MOLCA is discussed, and then the classical test function is used to test and analyze the MOLCA. Compared with NSGA-II, this method performs better than NSGA-II. MOLCA was applied to the optimization of the operation parameters of the main fractionator of FCC. With the total economic benefit and the energy consumption of the system as the two objectives, the optimized operation scheme was given. (4) in view of the problem that NSGA-II is easy to converge to the local optimal solution and the computation time is long, a hybrid algorithm MOLCA-NSGA-II. based on multi-objective queue competition algorithm and non-dominated genetic algorithm is proposed. The test results of classical test function show that the algorithm is superior to NSGA-II. in computing time, convergence and distribution. MOLCA-NSGA-II was applied to the optimization of the separation process of methanol to olefin. A series of optimal solutions of Pareto were given. According to different production requirements, energy consumption and yield could be considered synthetically, and suitable operating conditions could be selected.
【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ015.9

【参考文献】