Scholarlink

<p>To tackle the complex difficulties of recycling and remanufacturing end-of-life new energy vehicles (NEVs) and their components, this study develops a multi-channel, multi-cycle, multi-objective remanufacturing green supply chain network model under differentiated government intervention, and proposes an improved multi-objective deep reinforcement learning algorithm (IMO-DRLA) for its solution. The results indicate that: (1) the multi-channel, multi-cycle, multi-objective model greatly enhances overall supply chain optimization, enabling high-quality development; (2) by integrating fuzzy data preprocessing, deep learning–based prediction, and hyper-heuristic search for multi-objective optimization, the IMO-DRLA attains multi-objective optima in complex recycling scenarios, improving both computational efficiency and accuracy; and (3) differentiated subsidy and tax policies effectively strengthen supply chain stability. By precisely forecasting and solving the synergistic effects between the IMO-DRLA and differentiated governmental interventions, this study significantly improves the efficiency of multiple recycling channels for end-of-life NEVs, the emission-reduction rate, and the adoption of internet technologies, thereby providing critical methodological support for building a sustainable NEV recycling system.</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

Optimization of multi-cycle multi-objective remanufacturing green supply chain network based on improved deep reinforcement learning algorithm considering differentiated incentive compatibility mechanisms

  • 期刊论文

  • 发表日期：

    2026年3月2日

查看全文
  • Zhen Wang,
  • Chunming Ye,
  • Jianquan Guo

摘要

To tackle the complex difficulties of recycling and remanufacturing end-of-life new energy vehicles (NEVs) and their components, this study develops a multi-channel, multi-cycle, multi-objective remanufacturing green supply chain network model under differentiated government intervention, and proposes an improved multi-objective deep reinforcement learning algorithm (IMO-DRLA) for its solution. The results indicate that: (1) the multi-channel, multi-cycle, multi-objective model greatly enhances overall supply chain optimization, enabling high-quality development; (2) by integrating fuzzy data preprocessing, deep learning–based prediction, and hyper-heuristic search for multi-objective optimization, the IMO-DRLA attains multi-objective optima in complex recycling scenarios, improving both computational efficiency and accuracy; and (3) differentiated subsidy and tax policies effectively strengthen supply chain stability. By precisely forecasting and solving the synergistic effects between the IMO-DRLA and differentiated governmental interventions, this study significantly improves the efficiency of multiple recycling channels for end-of-life NEVs, the emission-reduction rate, and the adoption of internet technologies, thereby providing critical methodological support for building a sustainable NEV recycling system.