Scholarlink

<p>In view of steadily increasing waste volumes and material diversity in products, as well as increasing recycling requirements, existing collection, sorting, and recycling structures often reach their limits. Manual sorting, which is still used today, is costly, labor-intensive, and time-consuming, and is often uneconomical given the low economic value of many waste streams. Many sensor technologies have technical limitations (e.g. moisture or black objects in NIR) or are associated with very high costs and low throughputs (e.g. metal sorting using XRF or LIBS). At the plant level, individual processing and sorting units are mostly isolated solutions, and individual, short-term optimizations and coordination between the units for the respective waste stream are hardly possible. The Chair of Waste Processing Technology and Waste Management at the Montanuniversität Leoben is harnessing the high potential of artificial intelligence to meet current challenges in the industry in the best possible way, drawing on innovative, individually adaptable, and unique research equipment. This article presents ongoing projects in the fields of application of artificial intelligence and machine learning in waste management. Furthermore, a&#xa0;practical process for implementing an AI-based sorting model and the associated equipment is presented. In addition, the importance of representative data sets for efficient, generalizing, and robust AI in waste management is demonstrated.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Künstliche Intelligenz in der Abfallwirtschaft – Ein Überblick aktueller Anwendungen und zukünftiger Trends

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月21日

查看全文
  • Hannah Weber,
  • Julian Aberger,
  • Gerald Koinig,
  • Thomas Nigl,
  • Renato Sarc,
  • Alexia Tischberger-Aldrian

摘要

In view of steadily increasing waste volumes and material diversity in products, as well as increasing recycling requirements, existing collection, sorting, and recycling structures often reach their limits. Manual sorting, which is still used today, is costly, labor-intensive, and time-consuming, and is often uneconomical given the low economic value of many waste streams. Many sensor technologies have technical limitations (e.g. moisture or black objects in NIR) or are associated with very high costs and low throughputs (e.g. metal sorting using XRF or LIBS). At the plant level, individual processing and sorting units are mostly isolated solutions, and individual, short-term optimizations and coordination between the units for the respective waste stream are hardly possible. The Chair of Waste Processing Technology and Waste Management at the Montanuniversität Leoben is harnessing the high potential of artificial intelligence to meet current challenges in the industry in the best possible way, drawing on innovative, individually adaptable, and unique research equipment. This article presents ongoing projects in the fields of application of artificial intelligence and machine learning in waste management. Furthermore, a practical process for implementing an AI-based sorting model and the associated equipment is presented. In addition, the importance of representative data sets for efficient, generalizing, and robust AI in waste management is demonstrated.