Optimising microbial protein biotechnology using machine learning and mathematical optimisation for protein sustainability

使用机器学习和数学优化来优化微生物蛋白质生物技术以实现蛋白质可持续性

• 批准号: 2725958
• 金额: --
• 依托单位: King's College London
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2022
• 资助国家: 英国
• 起止时间: 2022 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2725958
• 关键词: Optimising; microbial; protein; biotechnology; using

This project draws upon cross-disciplinary expertise from King's College London (KCL) and well supported by Quorn Foods. Prof Rob Johnson from Quorn and Miao Guo from KCL have developed strong collaboration to drive exciting and timely research. They have been collaborating on lignocellulosic mycoprotein research and co-supervising 3 PhDs. The lab and modelling research in Guo's lab and Lam's intelligent control lab form the foundation for this programme.The plant-sourced and animal-sourced proteins are carbon-intensive, resource-demanding, and vulnerable to pandemic impacts due to long-production cycles. This combined with increasing protein demands and pandemic of hunger highlight the challenges on providing sustainable protein solutions. Microbial proteins produced under controlled fermentation with advanced optimisation- aided design enable rapid biotechnology advancement for future protein security and sustainability.This project aims to couple machine learning and advanced optimisation approach to enable step- change in microbial protein biotechnologies to achieve zero-waste, zero-emissions, where sustainable Mycoprotein innovation will be tested as a representative study. The research objectives (Obj) are -Obj-1 Novel integration of machine learning with model predictive control (MPC) to enable optimised and autonomous Mycoprotein biotechnology.This objective, underpinned by process control, optimisation theory and machine learning techniques, will develop a learning-based controller to harness real-time data and optimise mycoprotein fermentation to achieve maximised resource efficiency and minimised waste; such fermentation represents an advanced industrial biotechnology with multivariable dynamics, nonlinearities, and constraints. The data-driven machine learning techniques will play a significant role to deal with the highly-nonlinear and complex process of mycoprotein biotechnology that the analytical methods are difficult to be applied. The data-driven approach offers a feasibale alternative to reveal undercovered characteristics through machine learning. The developed learning-based MPC approach will advance microbial protein technologies and enable autonomous fermentation with waste recovery and optimised resource utilisation.Obj-2 Advanced optimiation to enable precision decision on net-zero Mycoprotein supply chains.Coupling data value chains and advanced computational modelling has the potential to bring tremendous opportunities to protein biotechnology to enable real-time data acquisition, analysis and data-informed responsive optimisation to achieve sustainability across supply chains. Underpinned by this new concept, Obj-2 will couple simulation, life cycle sustainability, mathematical optimiation and learning algorithms to develop a novel data-driven optimisation with hybrid solution search algorithms to bring real-time supply chain data (e.g. sensor data) to precision decision-support; the optimisation approach will be tested on Mycorpotein supply chains to achieve zero-emissions. Obj-2 approach is expected to catalyse protein sector transformation by shifting isolated technologies towards performance-optimised and machinery-interconnected ecosystem

该项目借鉴了伦敦国王学院（KCL）的跨学科专业知识，并得到了Quorn Foods的大力支持。Quorn的Rob Johnson教授和KCL的Miao Guo教授开展了强有力的合作，推动了令人兴奋和及时的研究。他们一直在木质纤维素真菌蛋白研究方面进行合作，并共同指导3个博士学位。郭的实验室和林的智能控制实验室的实验室和建模研究构成了这个项目的基础。植物来源和动物来源的蛋白质是碳密集型的，需要资源，并且由于生产周期长，容易受到大流行的影响。这与不断增加的蛋白质需求和饥饿的流行相结合，突出了提供可持续蛋白质解决方案的挑战。微生物蛋白生产控制发酵与先进的优化辅助设计，使生物技术的快速进步，为未来的蛋白质安全和可持续性。该项目旨在结合机器学习和先进的优化方法，使微生物蛋白生物技术的逐步变化实现零浪费，零排放，其中可持续的真菌蛋白创新将作为代表性研究进行测试。研究目标（Obj）是-Obj-1机器学习与模型预测控制（MPC）的新型集成，以实现优化和自主的真菌蛋白生物技术。这一目标以过程控制、优化理论和机器学习技术为基础，将开发一种基于学习的控制器，以利用实时数据和优化真菌蛋白发酵，实现资源效率最大化和浪费最小化；这种发酵代表了一种先进的工业生物技术，具有多变量动力学、非线性和约束。数据驱动的机器学习技术将在处理分析方法难以应用的高度非线性和复杂的真菌蛋白生物技术过程中发挥重要作用。数据驱动的方法提供了一个可行的替代方案，通过机器学习揭示隐藏的特征。开发的基于学习的MPC方法将推进微生物蛋白技术，并通过废物回收和优化资源利用实现自主发酵。Obj-2先进的优化，实现对净零真菌蛋白供应链的精确决策。将数据价值链和先进的计算建模相结合，有可能为蛋白质生物技术带来巨大的机会，使实时数据采集、分析和数据知情的响应优化成为可能，从而实现供应链的可持续性。在这一新概念的基础上，Obj-2将结合仿真、生命周期可持续性、数学优化和学习算法，开发一种新的数据驱动优化，采用混合解决方案搜索算法，将实时供应链数据（例如传感器数据）提供精确决策支持；优化方法将在菌蛋白供应链上进行测试，以实现零排放。Obj-2方法有望通过将孤立的技术转向性能优化和机械互联的生态系统来催化蛋白质部门的转型