追踪间歇过程演变的多变量统计过程监测及质量预测研究

Batch processes have been widely employed in chemical and other manufacturing industries, due to their capabilities to meet the requirements of fast changing markets and to manufacture small-lot and high-value-added products. The three-dimensional data structure of batch processes leads to the two-dimensional variation of process characteristic in the inner-batch direction and the inter-batch direction. In the inner-batch direction, many existing methods ignore the process variations between phases or within a phase. In the inter-batch direction, traditional methods handle process variations by adjusting the models consecutively, leading to the increase in probability of introducing disturbances and faults. .In this project, a typical kind of process characteristic variation caused by process dynamics or long term external factors is called process evolution, and both of the inner-batch evolution and the inter-batch evolution of batch processes will be studied. Several novel strategies will be proposed to improve both process monitoring and quality prediction for batch processes by tracing the inner-batch evolution and the inter-batch evolution: by analyzing the inner-batch evolution, data from uneven inner-phase parts are synchronized and proper models are built for transitions on the basis of their special characteristics; according to the inter-batch evolution, a whole batch process is divided into different modes, and the relationship between these modes is analyzed to dig up between-mode information; based on the analysis and modeling of batch process evolution, corresponding online monitoring strategy is proposed; taking the reality of batch processes into consideration, quality prediction strategy is presented; the platform for batch processes is developed for the validation and improvement of the proposed methods and for their application to typical industrial batch processes.

由于可以适应市场需求的迅速变化，生产小批量、高附加值产品的间歇生产方式在现代工业中普遍采用。三维数据结构决定了间歇过程的特性变化既体现在批次内也体现在批次间。在批次内，现有方法忽略了阶段之间的过渡过程以及阶段内部的变化；在批次间，传统方法持续调整过程模型，增加了引入扰动与故障的可能性。.本项目将由过程本身决定的或外部因素的长期作用而造成的典型的过程特性变化定义为过程演变，重点研究间歇过程批次内与批次间的过程演变，在追踪过程演变的基础上系统的提出改进间歇过程监测及质量预测的策略方法：基于批次内演变分析，根据过渡过程的特殊性建立适合的模型，合理地对不等长的阶段进行分部分等长；根据批次间演变特性分析进行模态划分，分析多模态关系，挖掘模态间信息；根据间歇过程演变的分析及建模，提出相应的在线监测策略；结合间歇过程生产实际，提出质量预测策略；开发应用测试平台，并面向典型间歇过程进行推广应用。

由于可以适应市场需求的迅速变化，生产小批量、高附加值产品的间歇生产方式在现代工业中普遍采用。三维数据结构决定了间歇过程的特性变化既体现在批次内也体现在批次间。本项目将由过程本身决定的或外部因素的长期作用而造成的典型的过程特性变化定义为过程演变，重点研究间歇过程批次内与批次间的过程演变，在追踪过程演变的基础上系统地提出改进间歇过程监测及质量预测的策略方法：首先，建立了追踪间歇过程批次内演变的模型：基于批次内演变分析，用潜在的演变来揭示一个批次进程发展的关键信息；通过批次内演变的分析与对应，建立多阶段之间关系；研究一般间歇过程的演变特点，创建多阶段质量残差递推模型和基于多阶段间关系分析模型两种模型进行模型训练，进而准确地捕捉多阶段批次过程特性。其次，建立了追踪间歇过程批次间演变的模型：基于过程演变分析进行批次间模态划分及模型建立，提出新模态产生的判别标准；分析模态与模态之间的演变关系，提取有价值的信息；基于模态间演变关系分析建立过程模型。再次，提出一套完整的追踪间歇过程演变的过程监测及质量预测策略：在线判断当前批次内演变情况以及当前批次间演变情况；基于演变分析，采用相应的监测模型对当前采样点进行监测；将演变分析与生产实际相结合，采用相应的质量预测模型预测产品质量。最后，进行工业仿真及实验验证。选择注塑成型过程作为对象搭建实际应用测试平台，对项目中提出的监测及质量预测算法进行工业应用测试和改进。