Supporting Analytics for Nuclear Asset Data Lifecycle

这项博士研究将为核工业开发可靠的分析工具，从工厂状态监测(CM)数据中提取可靠的信息。在严格监管的环境中运行的关键资产的预测和健康管理(PHM)是一项数据密集型和耗时的任务。核工业旨在通过以下方面的基础研究提高现有人工PHM方法的准确性和效率：新的知识和数据驱动的人工智能技术/过程，以及在混合系统(随机和基于物理的数字孪生模型)中提供多种信息源的集成手段。这项博士研究将询问是否可以研究创新的颠覆性技术，并将其应用于真实资产的CM数据，以预测资产退化并自动评估PHM数字供应链的完整性。这项研究将侧重于为工业应用创造和转化分析技术和人机界面方面的进步，以便核运营商能够自信地与遗留资产和未来资产中可用的丰富数据进行互动。技术挑战需要专门的博士深度调查，以跨越以前需要更长时间才能吸收新贡献的专属学科，例如：编码领域专家知识；基于物理的模型开发；或不可观察关系的机器学习(虚拟计量或替代传感器)。创新之处在于确定候选模型来验证代表性案例研究，例如：相关输入不明确的具有大量变量的物理模型；数字配对中的复杂关系；将不确定性编码为基于物理的模型。该博士将成为斯特拉斯克莱德大学高级核研究中心(ANRC)博士组合的一部分，他们是NPL的战略合作伙伴，并将把核供应链中的利益相关者聚集在一起，更好地整合他们的需求和要求。

This PhD research will develop trustworthy analytical tools for the nuclear industry, extracting reliable information from plant condition monitoring (CM) data. Prognostics and Health Management (PHM) for critical assets operating in strictly regulated environments is a data intensive and time consuming task. The nuclear industry aims to improve upon the accuracy and efficiency of existing manual PHM approaches, through fundamental research into: new knowledge and data-driven AI techniques/processes, and the means of integration offering multiple sources of information in hybrid systems (stochastic and physics-based digital twin models). This PhD research will ask if innovative, disruptive technologies can be investigated and applied to real assets' CM data, to predict asset degradation and automatically assess the integrity of the PHM digital supply chain. The research will focus on the creation and translation of advances in analytical techniques and human-machine interfaces for industrial applications, so that nuclear operators can confidently interact with the wealth of data available from legacy and future assets. The technical challenge requires dedicated PhD depth investigation to cut across previously exclusive disciplines that require a longer time to assimilate novel contributions, e.g.: encoding domain expert knowledge; physics-based model development; or, machine learning of unobservable relations (virtual metrology or surrogate sensors). Novelty is in identifying candidate models to validate representative case studies, e.g.: physics models with large numbers of variables where relevant inputs are unclear; complex relations in digital twinning; encoding uncertainty into physics-based models.The PhD will form part of the Advanced Nuclear Research Centre's (ANRC) PhD portfolio at the University of Strathclyde, who are an NPL strategic partner, and will bring together stakeholders in the nuclear supply chain to better integrate their needs and requirements.