Modelling and Control for the Advancement of Biologics

本研究计划的重点是建立一种新的先进的过程控制策略来操作进料间歇式生物反应器。该建议将利用人工智能方法来确定一个过程模型，该模型将允许更明智的控制方法。目前，批处理使用基于历史成功的固定协议进行操作。该行业面临的主要挑战之一是细胞对环境变化的敏感性，即使是轻微的偏差也可能导致丢弃一批。这种方法是专门设计的，通过在建模方法中利用现有的细胞生长知识来维持健康的细胞环境。控制器将明确地考虑干扰对细胞生长的影响，同时能够同时调整所有输入。为了建立这种方法，将进行试点实验以收集建模所需的数据，最终，将用所提出的方法控制规模系统。这项工作将产生一种神经网络建模方法和模型预测控制算法，旨在最大限度地提高中国仓鼠卵巢细胞的产量。这将使生物反应器的操作更加准确，减少对历史数据的依赖，从而给生物制药企业带来巨大的利益。

This research proposal is focused on building a novel advanced process control strategy to operate a fed-batch bioreactor. This proposal will utilize artificial intelligence methods to identify a process model that will allow for a more informed control approach. Currently, batches are operated using a fixed protocol based on historical successes. One of the main challenges in this industry is the sensitivity of cells to changes in the environment, where even slight deviations can result in a discarded batch. This approach is specifically designed to maintain a healthy cell environment by utilizing existing knowledge of cell growth in the modelling approach. The controller will explicitly account for the effects of disturbances to cell growth while being able to adjust all of the inputs simultaneously. To build this approach, pilot experiments will be conducted to collect data for modelling, and eventually, scale systems will be controlled with the proposed approach. This work will result in a neural network modelling approach and model predictive control algorithm designed to maximize the production of Chinese hamster ovary cells. It is expected to yield large benefits to the biopharmaceutical company by making bioreactor operation more accurate and less reliant on historical data.