应用带补偿的机会约束规划策略优化精馏过程的研究  

Distillation process optimization by the methodology of chance constrained programming with recourse

作　　者：张化杰[1] 孙力[1] 陆雪峰[1] 

机构地区：[1]大连理工大学化工学院,辽宁大连116023

出　　处：《计算机与应用化学》2010年第10期1425-1427,共3页Computers and Applied Chemistry

摘　　要：精馏过程优化存在许多不确定变量,其对优化结果的影响是不可忽略的,所以需要找到真实反映不确定变量的数学模型,找到最优的求解方法得到精确解,进而得到最优的优化结果。本研究通过分析随机规划策略和不确定因素的特性及其对精馏优化过程的影响,对不确定因素进行分类,应用带补偿的随机规划和机会约束规划实现不确定因素对优化目标和约束的影响。对于模型中的机会约束,由于不确定输入变量和不确定输出变量间存在单调关系,所以可以通过对不确定输入变量的多重积分将机会约束转化为等价地确定性非线性约束,此时模型转化为含有非线性约束的带补偿的随机规划模型,此模型通过改进的蒙特卡罗(MonteCarlo)积分和含有序列2次规划(SQP)的Benders分解相结合的混合算法有效的求解。所提出的求解方法借助于Matlab软件能够得到比较精确的结果。采用本文所提出的模型及求解方法对精馏塔操作的模拟实验过程进行优化研究,可以看出带补偿的机会约束规划模型能够很好地反映不确定变量对优化过程的影响。所以本文提出的模型和求解方法是有效可行的,且对于以后的不确定优化研究具有很好的指导意义。Distillation process optimization existsed many uncertain variables,the influence of the uncertainty variables on optimization results could not be neglected.It was need to find the mathematical model which could reflect the Uncertain variables truly,then solved the mathematical mode and got the best optimization results.The methodology of Chance constrained programming with Recourse was proposed in this paper basing on the stochastic programming strategy,the characteristics of the uncertainty variables and the influence on distillation process.There was a monotone relation between an output variable and one of the uncertaininput variables.So it was realistic that the probability of holding the output constraint can be computed by integration in the corresponding regions of the uncertain variables.This transformed the probability into the deterministic nonlinear constraint and the mathematical model was transformed into the two stages with compensation stochastic programming with the nonlinear constraint. For the solution on the the two stages with compensation stochastic programming,a hybrid algorithm of both Monte Carlo integration and improved Benders decomposition strategies with SQP was studied in this paper attempting solve the optimization model.A case study illustrates the feasibility of the proposed methodology.A mathematical model which could reflect the uncertainty variables primely in the optimization was proposed.It wae also feasible in algorithm.So this paper will have good reference value in the future optimization study.

关 键 词：随机规划 机会约束 补偿 混合算法 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]