Self-Learning Reactor Systems for Automated Development of Kinetic Models

研究背景：该项目建议将利兹和阿兹开发的自动化反应器平台与能够识别动力学模型的混合整数线性规划技术相结合，以创建一个真正自主的系统，用于评估和开发可扩展的过程模型。这将允许在加速的时间框架内生产药品。到目前为止，利兹使用自优化流量系统的工作已应用于统计优化和反馈算法。本项目旨在利用该反应器平台实现模型判别，并自动生成动力学模型之间的判别，从而生成能够转移到其他设备类型的过程模型。该项目有三个关键要素，每个要素都有自己的重大学术研究挑战和需要回答的问题：1 .通过多点采样和传感器的集成自动生成动力学剖面。将研究沿流动反应器长度连接的采样回路的使用，以快速生成整个动力学剖面，而不需要改变流速（因此混合）。2。生成可行的动力学模型和判别将使用混合整数线性规划（见http://dx.doi.org/10.1016/j.compchemeng.2016.04.019）。这些技术将集成到反应器平台中，以允许对动力学模型进行快速评估和区分。示范模型转移和放大。生成的模型的适用性将通过大型设备和不同设备类型的评估来证明。

Context of Research:This project proposes to couple the automated reactor platforms developed at Leeds and AZ to mixed integer linear programming techniques capable of kinetic model discrimination to create a truly autonomous system for the evaluation and development of scalable process models. These will allow manufacture of pharmaceuticals in accelerated timeframes.Thus far the work using self-optimising flow systems at Leeds has been applied to statistical optimisations and feedback algorithms. This project aims to utilise this reactor platform to enable model discrimination and generate automatically discriminate between kinetic models and thus generate a process model capable to being transferred to other equipment types. The Project: has three key elements, each with its own significant academic research challenges and questions to be answered: I. Automated generation of kinetic profiles via integration of multipoint sampling and sensors. The use of sampling loops connected along the length of a flow reactor to rapidly generate whole kinetic profiles without requiring changes in flow rate (and hence mixing) will be investigated. II. Generation of feasible kinetic models and discrimination will be performed using mixed integer linear programming (see http://dx.doi.org/10.1016/j.compchemeng.2016.04.019). These techniques will be integrated within the reactor platform to allow rapid evaluation and discrimination of kinetic models.III. Demonstration of model transfer and scale-up. Applicability of the generated models will be demonstrated via evaluation in large scale equipment and different equipment types.