Development of machine learning tools for the characterization and sorting of low level waste

核电运行产生的低水平废物（LLW）的储存和处置是一项艰巨的挑战，因为低水平废物量很大。LLW通常储存在指定的储存设施中；然而，继续建造新的储存设施来满足快速增长的LLW量是不可持续的，也不经济。作为一种替代方案，现有的存储容量可以通过LLW排序和分离来优化。然而，在商业实践中，LLW材料的表征和分类只能通过人工来完成，这是一种非常粗糙、低技术和劳动密集型的方式。目前还没有用于商业应用的自动LLW排序工具。本研究的目标是利用先进机器学习的力量，开发自动化工具，对安大略省核舰队运行的LLW进行表征和分类。这项研究将与劳伦斯能源合作伙伴合作进行，这是核能行业的创新者和领导者。本研究的预期结果是：1)对核废料处理、表征、分类和分离技术进行了全面的文献综述；2)开发交互式LLW数据库，跟踪LLW，有效管理LLW数据；3)开发基于机器学习的LLW表征和分类工具，用于处理LLW图像和预测LLW内容。这项工作将导致最先进的机器学习工具的发展，用于自动LLW表征和优化核废料分类。它将为日益增长的LLW量问题提供长期解决方案，并有助于提高加拿大核能工业作为世界领先者的能力。

The storage and disposal of Low Level Waste (LLW) from nuclear power operations is a daunting challenge due to the large LLW volumes. LLW are typically stored in designated storage facilities; however, continuing to build new storage facilities to meet the fast growing volumes of LLW is not sustainable or economic. As an alternative, the existing storage capacity can be optimized by LLW sorting and segregating. However, the characterization and sorting of LLW materials in commercial practices can only be done manually, in a very crude, low-technology, and labor-intensive way. There is no existing automatic LLW sorting tools for commercial applications. The goal of this research is to leverage the power of advanced machine learning and develop automatic tools for the characterization and sorting of LLW from Ontario's nuclear fleet operation. This research will be conducted in collaboration with Laurentis Energy Partners, which is an innovator and leader in the nuclear energy industry. The anticipated outcomes of this research are: 1) a comprehensive literature review on nuclear waste handling, characterization, sorting, and segregation technologies; 2) development of an interactive LLW database for tracking LLW and effectively managing LLW data; and 3) development of machine-learning-based LLW characterization and sorting tools for processing LLW images and predicting LLW content. This work will lead to the development of state-of-the-art machine learning tools for automatic LLW characterization and optimized nuclear waste sorting. It will provide a long-term solution to the growing issue of LLW volumes and help enhance the capacity of Canadian nuclear energy industry as a world leader.