不确定条件下多产品间歇过程优化研究

Study on Optimization of Multiproduct Batch Process Under Uncertainty

针对不确定参数数目增大时不确定条件下间歇过程优化较难求解的情况 ,提出了新的求解策略。将蒙特卡罗积分策略与求解二阶段随机规划的可行域算法相结合 ;依据间歇过程特性 ,避免求解一系列可行域限定子问题 ;将积分值抽样点数与Benders算法主问题优化限定条件中的抽样点数相区别 ,利用两组抽样点求解。新的求解策略使不确定参数数目增大的情况下间歇过程优化问题变得较易处理 ,算例证明了该算法的有效性。

It is more difficult to save optimal design model of batch process under uncertainty with the increase of the number of uncertainty. For dealing with the puzzle easily, a new algorithm for multiproduct batch process optimization under uncertainty was proposed. The new approach was of three features combining Monte Carlo sampling with Feasible Region Algorithm proposed by Ierapetritou & Pistikopoulos(1995) to handle 2-stages stochastic program, avoiding the solution of feasibility subproblems according to the property of batch process design under uncertainty, two groups sampling numbers are differentiating, one for Monte Carlo integration and one for Lagrangian cut of Benders decomposition. The feasibility of the proposed algorithm was demonstrated with examples of multiproduct batch process design previously suggested in the open literatures.