基于增量式拟进化规划算法的TE模式切换最小时间动态优化

Minimum Time Dynamic Optimization of Grade Transition in TE Process Using a Novel Incremental Evolutionary Programming

以往对聚烯烃牌号切换的优化一般采用序列二次规划和迭代动态规划方法。本文针对不同成分产品之间的切换问题,建立两层控制结构的牌号切换系统:回路控制层和牌号切换层,通过控制变量参数方法转化为带有约束的非线性规划问题。针对该问题控制变量轨迹的特性,提出一种增量式的拟进化规划方法(IEA)。对于最小时间优化问题,采用双层优化策略:内层以某个给定末端时间产生可行解,搜索可行域;根据内层可行域的大小,收缩末端时间。将该方法应用于TE仿真系统,得到TE模型两种不同生产状态之间切换的最小时间和切换过程中操作变量的轨迹,验证了所提方法的有效性。

So far, many studies focus on the grade transition of polyolefin, and polyolefin grade transition optimization was solved by Sequential Quadratic Programming （SQP） or Iterative Dynamic Programming （IDP） traditionally. In this paper,to solve the problem of grade transition, we build a two layers of grade transition control structure . The two layers are loop control layer and grade transition layer. It is transformed into a constrained nonlinear programming problem by control variable parameters. We propose a novel Incremental Evolutionary Algorithm （IEA） which is similar to the evolutionary programming to solve this problem. Meanwhile, a Double-Layer Optimization Algorithm （DLOA） is adapted to solve the minimum time dynamic optimization problem. In DLOA, the inner optimization is to search the feasible region in the given final time （t~）; and in the outer layer, tf is reduced according to certain laws; finally, we get minimal tf and optimal trajectories of control variables.