面向大功率多微波源加热过程的机理-数据融合建模及协同优化

One of the keys to ensuring safety and efficiency in the industrial application of the high power microwave energy is that microwave power can be controlled based on he properties of the heated media in real-time. However, the most pressing obstacle is the lack of the appropriate mathematical model supported to design controller and the method of collaborative optimization in the microwave heating process with multi-sources for attaining real time control of microwave power. The main research items of this project are as follows..(1) We will study the mechanism of microwave heating and the fast resolution of spatial-temporal decouple and dimensionality reduction based on ‘full’ media properties, and clarify the role of relationship of ‘Microwave-Time-Temperature-Mass’ in microwave heating process. Furthermore, we will establish mathematics model of heat-mass and electromagnetic that not only can be solved faster but also benefits designing a controller of the microwave power..(2) We will study the fusion method based on mechanism of microwave heating and bounded data of microwave heating process, and establish ‘mechanism and data’ fusion model for describing the media temperature distribution in the practical microwave heating process. .(3)We will study the spatial and temporal correlation of microwave, media, operating index and condition, and their role in microwave distributed heating system. Meanwhile, we will establish the method of collaborative optimization in the microwave heating process with multi-sources and achieve safe and efficient heating goal in the end..(4) The proposed theory and method will be applied and validated in the 15 KW tunnel microwave heating experiment system for solving some critical problems in microwave heating process..This project will contribute to theoretical basis and theoretical guarantee for the safety and efficiency in the development and industrial application of microwave energy.

基于媒质特性实时调控微波输出功率是保障大功率微波能“安全和高效”工业应用的关键之一。然而，缺乏适用于控制器设计的数学模型和多微波源的协同机制是实现多微波源功率实时调控的主要障碍。本项目研究基于“完全”媒质特性的微波加热机理及其时空解耦-降维快速求解方法，阐明“微波-时间-温度-质量”关系，构建适用于快速求解和控制器设计的“热质-电磁”耦合模型；利用有限数据信息及先验知识，研究机理与数据融合方法，构建“机理-数据”融合模型以适用于具有混杂工况的微波加热过程；研究多微波源分布式加热系统中微波源与期望性能在时空域中相互作用规律，建立多微波源分布式加热过程中的协同机制与优化调控方法，完成“安全和高效”的加热目标。本项目将理论研究与微波能工业应用实际需求紧密结合，在15 KW大功率隧道式微波加热实验系统进行应用验证。本项目的研究成果将为微波能“安全和高效”工业应用提供理论和技术保证。

为了实现工业级微波能的安全、高效及可靠应用，本课题围绕大功率隧道式微波加热过程中“建模”和“控制”两大关键性问题展开一系列研究。微波加热过程的机理模型可用来描述电磁场与热动力学场的自身演变及相互耦合关系，通常由多个高维偏微分方程组成。从控制的角度来讲，这类机理模型阶数高、非线性强，很难基于此设计实时有效的控制器。尤其是对于工业级大规模隧道式微波加热系统，多个微波源分布式馈入加热腔体，且媒质随时间发生空间位移。不断变化的幅相界面和运行工况使得电磁场及温度场的时空分布及演化过程变得更加复杂，建立机理模型几乎变得不可能。数据模型因此成为一种有效的替代工具。数据驱动方法不依赖于受控过程的模型信息，仅利用在线或离线数据便能实现过程的建模与控制。因此，为了保障隧道式微波加热过程的高效安全运行，需要基于数据驱动的思想，建立数据模型对系统特性进行准确辨识，以实现高效智能控制。为此本课题的主要研究工作和成果包括如下几个方面：1）更新了15kW隧道式微波加热智能控制系统，该系统可用于过程数据的采集和微波源的实时调控，所搭建平台是精确、实时控制大功率微波源系统工作的重要物理基础；2）基于海量丰富的传感信息和过程数据，研究数据驱动的隧道式微波加热过程建模方法，分别设计了NARX神经网络和递归神经网络温度预测模型，该模型是控制器设计的基础；3）由于时变的媒质参数，微波加热过程通常呈强时变性和非平稳性，因此进一步研究模型的在线更新机制，并分别提出基于多局部模型集成和全局模型学习的在线建模方法；4）最终基于离、在线数据和过程模型，研究多微波源协同优化控制策略，分别提出分层递阶控制、基于自适应动态规划和事件触发的最优跟踪控制以及基于COMSOL数值模拟的多微波源控制策略。