Scholarlink

<p>The transformation of local public transportation towards sustainable technologies is essential for achieving sustainability goals, such as those outlined by the UN and national strategies. The <i>European Green Deal</i> compels this in the EU and mandates that all public transport operators have to transition their vehicle fleets to sustainable propulsion technologies by 2050. This transition primarily focuses on procuring and deploying electric buses (e-buses), balancing cost and sustainability considerations. However, the operation and usage of e‑buses deviate from traditional diesel buses due to their limited range, which is influenced by numerous factors. Initial adaptive approaches based on Machine Learning (ML) to predict the energy consumption of e‑buses have been explored in academia. Nevertheless, these remain largely theoretical and lack practical insights incorporating operational data and processes. This contribution addresses this gap and presents the results of a&#xa0;collaboration with Hamburger Hochbahn AG. Employing a&#xa0;practical and design-oriented approach, we used and analyzed operational data, conducted fifteen expert interviews, and trained several ML models with an average deviation of below 0,4% to predict the energy consumption of e‑buses. Our findings outline multiple challenges and eight derived design principles that serve as recommendations for developing prediction systems for e‑buses to support the transformation toward sustainable mobility.</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

Entwicklung von ML-basierten Vorhersagesystemen für den Energieverbrauch von Elektrobussen im öffentlichen Nahverkehr

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月21日

查看全文
  • Marten Borchers,
  • Hauke Volquardsen,
  • Jan Willruth,
  • Enrico Milutzki,
  • Martin Semmann,
  • Eva Bittner

摘要

The transformation of local public transportation towards sustainable technologies is essential for achieving sustainability goals, such as those outlined by the UN and national strategies. The European Green Deal compels this in the EU and mandates that all public transport operators have to transition their vehicle fleets to sustainable propulsion technologies by 2050. This transition primarily focuses on procuring and deploying electric buses (e-buses), balancing cost and sustainability considerations. However, the operation and usage of e‑buses deviate from traditional diesel buses due to their limited range, which is influenced by numerous factors. Initial adaptive approaches based on Machine Learning (ML) to predict the energy consumption of e‑buses have been explored in academia. Nevertheless, these remain largely theoretical and lack practical insights incorporating operational data and processes. This contribution addresses this gap and presents the results of a collaboration with Hamburger Hochbahn AG. Employing a practical and design-oriented approach, we used and analyzed operational data, conducted fifteen expert interviews, and trained several ML models with an average deviation of below 0,4% to predict the energy consumption of e‑buses. Our findings outline multiple challenges and eight derived design principles that serve as recommendations for developing prediction systems for e‑buses to support the transformation toward sustainable mobility.