Scholarlink

Global energy transition drives rapid development in automotive powertrain technologies, generating massive knowledge across electric, hybrid, and fuel cell vehicles. Traditional knowledge management methods struggle with multi-source heterogeneous data, creating knowledge islands. This study proposes a comprehensive knowledge graph construction method for automotive powertrain energy knowledge management. Our approach includes: (1) a domain-specific knowledge extraction framework for multi-source heterogeneous data; (2) a knowledge fusion mechanism resolving conflicts and redundancy; (3) graph neural network validation for discovering hidden relationships; (4) isolated node analysis. Experimental results demonstrate our knowledge graph contains 1,579 entities and 1,932 relationships, successfully revealing complex multi-hop chains and identifying key hub nodes in technical competition landscapes.

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

LLM-Driven Knowledge Discovery and Fusion for Automotive Powertrain Energy Systems

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yuanyuan Li,
  • Hailiang Zhang,
  • Jing Cao,
  • Jialin Rong,
  • Jiazi Ouyang,
  • Yumeng Zhu

摘要

Global energy transition drives rapid development in automotive powertrain technologies, generating massive knowledge across electric, hybrid, and fuel cell vehicles. Traditional knowledge management methods struggle with multi-source heterogeneous data, creating knowledge islands. This study proposes a comprehensive knowledge graph construction method for automotive powertrain energy knowledge management. Our approach includes: (1) a domain-specific knowledge extraction framework for multi-source heterogeneous data; (2) a knowledge fusion mechanism resolving conflicts and redundancy; (3) graph neural network validation for discovering hidden relationships; (4) isolated node analysis. Experimental results demonstrate our knowledge graph contains 1,579 entities and 1,932 relationships, successfully revealing complex multi-hop chains and identifying key hub nodes in technical competition landscapes.