Scholarlink

<p>Previous studies have employed hybrid heuristic algorithms (such as the improved pigeon flock algorithm and genetic algorithm) to address the high mortality rate of nodes, prolonged charging delays and high energy consumption of mobile chargers (MCs) in wireless rechargeable sensor networks (WRSNs). However, these algorithms are either designed for static scenarios or perform poorly when scaled up. Furthermore, some dynamic optimisation models (e.g. high-dimensional multi-objective models) have poor convergence efficiency. Traditional DQN struggles to balance charging efficiency and network lifetime due to overestimating Q-values and lacking clustering integration. To overcome these limitations, this study proposes an integrated strategy comprising DDQN, dynamic energy thresholds and K-means clustering. K-means clustering enables batch charging to reduce overheads; dynamic thresholds prevent energy depletion or waste; and DDQN’s dual-network architecture decouples optimisation sequences. Simulation results show that, compared with algorithms such as FCFS and NJNP, this strategy can significantly reduce node mortality (by 18–45%), charging delays (by 22–50%) and the travel distance of mobile chargers (by 25–40%).</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

Research on charging strategies for wireless rechargeable sensor networks based on dual-depth Q networks

  • 期刊论文

  • 发表日期：

    2026年2月2日

查看全文
  • Tian Peng,
  • Yang Jia,
  • Pu Hongyu,
  • Tian Xin,
  • Ran Guozheng,
  • Peng Liang

摘要

Previous studies have employed hybrid heuristic algorithms (such as the improved pigeon flock algorithm and genetic algorithm) to address the high mortality rate of nodes, prolonged charging delays and high energy consumption of mobile chargers (MCs) in wireless rechargeable sensor networks (WRSNs). However, these algorithms are either designed for static scenarios or perform poorly when scaled up. Furthermore, some dynamic optimisation models (e.g. high-dimensional multi-objective models) have poor convergence efficiency. Traditional DQN struggles to balance charging efficiency and network lifetime due to overestimating Q-values and lacking clustering integration. To overcome these limitations, this study proposes an integrated strategy comprising DDQN, dynamic energy thresholds and K-means clustering. K-means clustering enables batch charging to reduce overheads; dynamic thresholds prevent energy depletion or waste; and DDQN’s dual-network architecture decouples optimisation sequences. Simulation results show that, compared with algorithms such as FCFS and NJNP, this strategy can significantly reduce node mortality (by 18–45%), charging delays (by 22–50%) and the travel distance of mobile chargers (by 25–40%).