Automatic quality assessment of waste plastic bales through hybrid sensing and data driven modelling

通过混合传感和数据驱动建模对废塑料包进行自动质量评估

• 批准号: EP/W026228/1
• 负责人: Lijuan Wang
• 金额: $ 51.1万
• 依托单位: University of Kent
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FW026228%2F1
• 关键词: Automatic; quality; assessment; waste; plastic

Plastic waste is one of the most serious environmental challenges across the world. Currently, around 381 million tonnes of plastic waste are generated every year in the worldwide. However, the current recycling rates are thought to be 14%-18% at the global level. A key issue causing the low recycling rate is the uncertainty of the quality of recycled plastics. It is contaminated with a wide range of non-plastic and non-recycling materials. The current method for inspection of baled materials is core sampling. However, this method is slow to determine results and not accurate if samples extracted are not representative. Therefore, an automatic and non-destructive measurement technology is highly desirable for accurate identification and quantification of materiel type and composition in waste plastic bales and further assessment of the quality of bales.This project proposes a new measurement methodology for automatic quality assessment of waste plastic bales through hybrid sensing and data driven modelling. The project will start with design and construction of a hybrid sensing unit including a hyperspectral scanner and a 3D capacitive sensor. Meanwhile, considerations about the optimal design of the capacitive sensor with multiple electrodes will be made through finite element modelling. Identification and quantification algorithms based on machine learning and 3D reconstruction techniques will be developed to visualise material distribution and identify material compositions. Quality assessment algorithms based on expert knowledge and artificial intelligence will be developed to classify the plastic waste bale under test into a quality category. At the end of the project, the effectiveness of the proposed measurement methodology will be evaluated through laboratory tests and demonstration trials. The proposed methodology will provide a new way for inspection of baled materials from both surface and interior of the bale. Meanwhile, this measurement methodology makes the assessment automatic, which will significantly improve the efficiency of recycling processes. In addition, the established assessment criteria will ensure a clearer and more accurate definition of the quality of materials, which will result in more transparent pricing and greater clarity in global trade of waste plastics.

塑料垃圾是世界上最严重的环境挑战之一。目前，全球每年产生约3.81亿吨塑料垃圾。然而，在全球范围内，目前的回收利用率被认为是14%-18%。造成回收利用率低的一个关键问题是回收塑料质量的不确定性。它被各种各样的非塑料和非回收材料污染。目前对捆扎材料的检验方法是岩心抽样。然而，如果提取的样品不具有代表性，这种方法确定结果的速度很慢，而且不准确。因此，为了准确识别和量化废旧塑料包中的材料类型和成分，并进一步评估废旧塑料包的质量，迫切需要一种自动、无损的测量技术。本项目提出了一种新的测量方法，通过混合传感和数据驱动建模来实现废旧塑料包质量的自动评估。该项目将首先设计和建造一个混合传感单元，包括一个高光谱扫描仪和一个3D电容式传感器。同时，通过有限元模型对多电极电容式传感器的优化设计进行了考虑。将开发基于机器学习和3D重建技术的识别和量化算法，以可视化材料分布和识别材料成分。将开发基于专家知识和人工智能的质量评估算法，将测试中的塑料垃圾包归类为质量类别。在项目结束时，将通过实验室测试和示范试验来评估拟议测量方法的有效性。该方法将为捆扎材料的表面和内部检验提供一种新的方法。同时，这种测量方法使评估自动化，这将显著提高回收过程的效率。此外，既定的评估标准将确保对材料质量有更明确和更准确的定义，从而使全球废塑料贸易的定价更加透明，更加透明。