自适应粒子群法于应用催化裂化分馏塔多目标的优化  

The application of adaptively adjustment partical swarm optimization algrithm in catalyzing & & cracking fractionating tower

作　　者：刘蓉[1] 吕翠英[1] 

机构地区：[1]华南理工大学理学院,广东广州510640

出　　处：《计算机与应用化学》2010年第6期771-774,共4页Computers and Applied Chemistry

摘　　要：为了更有效地解决催化裂化分馏塔多目标优化问题,本文根据熊俊文等所建催化裂化分馏塔多目标优化问题的数学模型,提出采用自适应粒子群法作为优化手段。在传统粒子群算法中,由于每代粒子飞行时间固定为1,导致产生"振荡",且惯性权重ω是线性递减的,没有充分利用目标函数所提供的其它信息,使得搜索方向的启发性不强,收敛速度较慢且易陷入局部极值。本算法在传统粒子群法基础上,采取自适应调整飞行时间,减少"振荡"现象的产生,并且充分利用目标函数提供的信息,动态变惯性权值,增强算法的启发性,有效地跳出局部极值。通过参数调试,选取一组合适的参数优化分馏塔。将本法与申慧敏等IPAGA、周晓静等ASACA的优化结果及优化过程加以比较,结果表明:本法优化结果较好而且速度快,与ASACA最优解相同,无论综合测评函数还是子目标函数的取值均高于IPAGA,同时克服了ASACA中暂停现象,大大地减少了时间花费,有利于提高生产效率,是行之有效、多目标优化分馏塔的方法。但是本法在处理复杂的多目标优化问题时存在一定的局限性,还有待进一步改进。In order to solve the catalyzing cracking fractionating tower more effectively,this paper adopts the adaptively adjustment partical swarm optimization algrithm as intelligent optimization means which based on prior literatures which proposed the model of catalyzing cracking fractionating tower multi-objective problem.In the traditional partical swarm optimization algorithm,owing to lots of oscillation caused by the fixed flight-time of partical swarm on 1.And due to linearly reducing inertia-weight and less information about object function,these make the algorithm less inspiring on the searching direction.Then convergence slowly and easily fall into local maximum.This algorithm is on the base of traditional partical swarm optimization algorithm,and it adopts adaptively adjustment flight-time to reduce oscillation.It also makes full use of information of object-function,change the inertia-weight adaptivelywhich improves inspirement of algorithm,and avoid the local maximum efficiently.Then it finds out a group of suitable values by debugging the parameters in the process of researching.Compared with algorithm IPAGA in literature which proposed by Shen Huimin and algorithm ASACA in literature which proposed by Zhou Xiaojing,experiment shows that：the result of adaptively adjustment partical swarm optimization algrithm is better and speed of optimization is quicker.It makes the same maximum as ASACA which is higher than IPAGA on both comprehensive and sub objective function.At the same time the time-cost is reduced,which is good for effiency improvement,so it is feasible for the catalyzing cracking fractionating tower.But this algorithm has some limitations in dealing with comprehensive problems,still needs further improvement.

关 键 词：分馏塔 多目标优化 自适应 粒子群算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]