Metal recovery and recycling using machine learning and eddy current inspection

在减少浪费、节约有价值的原材料和最大限度减少温室气体排放的当务之急的推动下，全球工业正在转向材料和产品供应的循环经济模式。金属非常适合循环使用，这一点得到了成熟的金属回收行业的支持。然而，回收商在从混合金属废流(称为Zorba)中分离不同的非铁素体金属方面面临着巨大的挑战。在发展中国家，这些垃圾通常是手工分类的，或者是通过昂贵的、可能破坏环境的浮动水槽系统进行分类的。这种情况在经济和伦理上都是不可持续的，需要一种经济高效的干燥技术解决方案。这个博士项目将研究使用机器学习算法、计算机视觉和磁感应光谱分析对Zorba废料中的金属进行分类的新方法。该项目将进行新传感器的研究、设计和实施，并探索感应测量和碎片物理几何的新功能集和分析，以创建高效且经济实惠的解决方案，以提高回收率和纯度。该博士项目响应工业需求和UKRI战略。它跨越了EPSRC的“可持续工业”优先事项、IUK Horizons工具包确定的“浪费”和“不可再生资源”挑战等优先事项，并根据繁荣成果“高效和可持续地管理资源”的雄心。

Industry globally is moved to a circular economy model for the supply of materials and products driven by the imperative to reduce waste, conserve valuable raw materials and minimise green-house gas emissions. Metals are very well suited to the cyclic use and this is supported by a mature metal recycling industry. However, the recyclers face a significant challenge in separating the different non-ferritic metals from mixed-metal waste streams (known as ZORBA). These have typically been sorted by hand in developing countries or by expensive and potentially environmentally damaging float-sink systems. The situation is unsustainable both financially and ethically and a cost-effective dry technological solution is needed.This PhD project will investigate new approaches to classify metals in ZORBA waste streams using machine learning algorithms, computer vision and magnetic induction spectroscopy. The project will conduct research, design, and implement new sensors, and explore new feature-sets and analysis of induction measurements and fragment physical geometry to create efficient and cost-effective solutions to improve recovery and purity rates.This PhD project is responsive to industrial needs and UKRI strategies. It cuts across priorities such as the EPSRC 'Sustainable Industries' priority, 'Waste' and 'Non-renewable resources' challenges identified by the IUK Horizons toolkit, and 'Manage resources efficiently and sustainably' ambitions under the prosperity outcomes.