批次过程迭代建模与在线优化方法研究

发布时间：2018-04-18 06:10

本文选题：批次过程 + 质量控制　；参考：《浙江大学》2015年硕士论文

【摘要】：基于过程模型的优化方法是批次过程质量控制与操作优化的主流方法,然而其往往存在建模困难的问题；无模型优化方法是近阶段学者提出的优化方法,然而其存在在线实验成本高的问题。本文结合一类批次过程快速、低成本、可重复性的特性,并借鉴基于过程模型的优化方法与无模型优化方法各自的优点,提出了一种迭代建模与信赖域优化的方法,即局部建模、信赖域优化、在线试验相结合的方法。其具有较高的优化效率和较低的实验成本,一定程度上解决了建模过程中的模型失配问题,并且具有一定的收敛性。本文的主要研究内容总结如下：迭代建模与信赖域优化方法的关键之一是采用局部模型用于优化,本文提出一种迭代建模的策略,确保局部模型的精度逐步得到提高；其主要包括模型结构选择、数据集更新和模型残差校验。另一个关键点是信赖域优化,本文利用信赖域优化实现在优化问题求解中对优化步长的限制,确保迭代点的可信度和有效性。本文通过批次过程仿真算例验证了迭代建模与信赖域优化方法的可行性和有效性。其中对两类典型的优化问题进行了探讨,分别为目标函数未知的批次过程仿真优化和约束函数未知的批次过程仿真优化。通过对迭代过程的展示和分析,详细地介绍了该方法的求解过程和结果,验证了方法在求解不同优化命题中的可行性。本文还将此方法应用于实际工业过程中一类典型的批次过程——注塑成型过程。同样对两种典型的优化问题进行了探讨,分别为目标函数未知和约束函数未知,即过程模型体现在目标函数和约束函数中。实验结果证明了方法的可行性和有效性。本文还将该方法与传统方法进行了对比,实验结果有力地验证了该方法能更快更好地解决批次过程质量控制与操作优化问题。
[Abstract]:The optimization method based on process model is the mainstream method of batch process quality control and operation optimization.However, it has the problem of high cost of online experiment.In this paper, a method of iterative modeling and trust region optimization is proposed, which combines the characteristics of a class of batch processes such as fast, low cost and repeatable, and uses the advantages of process model-based optimization method and model-free optimization method for reference.Local modeling, trust region optimization and online test are combined.It has higher optimization efficiency and lower experimental cost, solves the model mismatch problem in the modeling process to a certain extent, and has certain convergence.The main research contents of this paper are summarized as follows: one of the keys of iterative modeling and trust region optimization is to use local model for optimization. In this paper, an iterative modeling strategy is proposed to ensure that the precision of local model is improved step by step;It mainly includes model structure selection, data set update and model residuals check.Another key point is trust region optimization. In this paper, we use trust region optimization to limit the step size of the optimization problem and ensure the reliability and validity of the iteration point.The feasibility and effectiveness of the iterative modeling and trust region optimization methods are verified by batch process simulation examples in this paper.Two kinds of typical optimization problems are discussed, one is batch process simulation optimization with unknown objective function and the other is batch process simulation optimization with unknown constraint function.Through the presentation and analysis of the iterative process, the solution process and results of the method are introduced in detail, and the feasibility of the method in solving different optimization propositions is verified.This method is also applied to a kind of typical batch process-injection molding process.At the same time, two typical optimization problems are discussed, which are unknown objective function and unknown constraint function, that is, the process model is embodied in the objective function and the constraint function.The experimental results show that the method is feasible and effective.The experimental results show that the method can solve the problem of batch process quality control and operation optimization more quickly and better.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ019;TP273

【参考文献】