基于过程数据的生产设计研究

发布时间：2018-05-15 00:41

本文选题：生产设计 + 隐变量模型　；参考：《东北大学》2014年硕士论文

【摘要】：在工业生产中,生产者通过改变工业生产过程的操作条件或原料的配比可以生产出不同等级的产品,以满足不同的市场需求。然而,如何根据客户期望的性能指标来寻找或调整相应的过程操作条件已逐渐成为一个普遍性问题,生产设计理论及方法是解决该问题的有效途径,尤其针对那些原材料成本高、生产周期长的产品的生产,如果能找到合适的生产设计方法来确定过程所需的操作条件,那么就可以有效缩短生产周期,降低生产成本,进而提高生产效率。因此,对于生产设计的研究具有理论及实际的意义。生产设计问题需涉及到过程模型的建立,本文首先介绍了几类常用的建模方法,其中基于过程数据建立的回归模型具有建模速度快、模型的内部结构透明、不需过多了解过程等特点,受到了广泛地关注及研究。然后,针对标准回归模型中回归参数由于操作变量的共线性而失效的问题,一类基于隐变量模型的方法(本文主要研究了PCR及PLS)被应用到生产设计中。最后,将上述的方法应用到一个低密度聚乙烯生产过程中进行仿真验证,仿真结果证明了隐变量模型求解生产设计问题的有效性。此外,对于过程数据信息不足时的生产设计问题,本文以相似生产过程的信息迁移策略研究为主线,对增广主元回归(EPCR)和联合输出偏最小二乘(JYPLS)两种方法进行了理论分析,并通过仿真分析说明了这两种方法求解生产设计问题的实用性和有效性。针对时变性过程的生产设计问题,本文采用了递推PCR算法,利用生产过程产生新数据不断地对过程模型进行更新；针对随着建模样本点逐渐增多而导致的更新模型适应过程动态变化特性变差的问题,本文采用了移动窗口策略来减弱历史数据对当前过程模型的影响。最后仿真实验说本文提出的生产设计方案在解决时变性的生产设计问题上的有效性。
[Abstract]:In industrial production, producers can produce different grades of products by changing the operating conditions of the industrial production process or the proportion of raw materials, in order to meet different market demands. However, how to find or adjust the corresponding process operating conditions according to the performance index expected by the customer has gradually become a universal problem. The production design theory and method is an effective way to solve this problem. Especially for the production of products with high raw material cost and long production cycle, if we can find a suitable production design method to determine the operating conditions required for the process, then we can effectively shorten the production cycle and reduce the production cost. And then improve the production efficiency. Therefore, the study of production design has theoretical and practical significance. The problem of production design involves the establishment of process model. Firstly, several kinds of modeling methods are introduced in this paper. The regression model based on process data has the advantages of fast modeling speed and transparent internal structure. Do not need too much understanding of the process and other characteristics, has received extensive attention and research. Then, aiming at the problem that the regression parameters in the standard regression model fail due to the collinearity of the operating variables, a class of implicit variable model (PCR and PCR) methods are applied to the production design. Finally, the simulation results show that the implicit variable model is effective in solving the production design problem of low density polyethylene (LDPE). In addition, for the problem of production design with insufficient process data information, this paper analyzes the two methods of augmented principal component regression (EPCR) and joint output partial least squares (JYPLS) based on the study of information migration strategy of similar production processes. The practicability and effectiveness of these two methods for solving production design problems are illustrated by simulation analysis. In order to solve the problem of production design of time-varying process, the recursive PCR algorithm is adopted in this paper, and the process model is updated continuously by using the new data generated by the production process. In order to solve the problem that the dynamic variation of updating model becomes worse due to the increasing number of sample points, a moving window strategy is used to reduce the influence of historical data on the current process model. Finally, simulation experiments show the effectiveness of the proposed production design scheme in solving the problem of time-varying production design.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TB49

【相似文献】