Scholarlink

Accurate estimation of the remaining operational lifespan (RoL) of lithium-ion batteries (LiBs) is essential for enhancing the safety, reliability, and performance of unmanned aerial vehicles (UAVs). Effective RoL prediction enables proactive battery management, minimizes the risk of mid-flight power failures, and supports optimal mission planning and scheduling. This study proposes a novel RoL estimation framework that employs an optimal feature selection method using the informative distance genetic algorithm (IDGA) and long short-term memory (LSTM) networks, by examining various measurable inputs including charging from the battery management system for addressing capacity regeneration. The subset of most aging-related features (MARF), derived from charging parameters, is integrated into a precisely calibrated network architecture. Comparative evaluations across various battery datasets demonstrate that the proposed method overcomes current benchmarks, consistently yielding precise RoL forecasts. The results validate the effectiveness and robustness of the proposed model, demonstrating significant potential for implementation in battery health management applications.

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

Estimating the RoL of UAV Lithium-Ion Battery Using IDGA-Based Feature Selection and LSTM Network Architecture

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Cong Dai Nguyen,
  • Van Khoe Ta

摘要

Accurate estimation of the remaining operational lifespan (RoL) of lithium-ion batteries (LiBs) is essential for enhancing the safety, reliability, and performance of unmanned aerial vehicles (UAVs). Effective RoL prediction enables proactive battery management, minimizes the risk of mid-flight power failures, and supports optimal mission planning and scheduling. This study proposes a novel RoL estimation framework that employs an optimal feature selection method using the informative distance genetic algorithm (IDGA) and long short-term memory (LSTM) networks, by examining various measurable inputs including charging from the battery management system for addressing capacity regeneration. The subset of most aging-related features (MARF), derived from charging parameters, is integrated into a precisely calibrated network architecture. Comparative evaluations across various battery datasets demonstrate that the proposed method overcomes current benchmarks, consistently yielding precise RoL forecasts. The results validate the effectiveness and robustness of the proposed model, demonstrating significant potential for implementation in battery health management applications.