基于强化学习的多无人机WRSN充电路径规划方案

Multi-UAV Collaborative Charging Path Planning in Wireless Rechargeable Sensor Network

针对无线可充电传感网(WRSNs)存在的充电难、效率低等问题,提出了一种多无人机协同无线可充电传感网充电路径规划方案(MC-CPP).首先描述了多无人机路径规划问题并建立了数学模型,然后针对问题模型提出了相应的深度强化学习(DRL)算法,算法借助了神经网络、贪婪策略和经验回放等获取无人机的充电飞行路径;最后无人机沿着规划的路径为网络中各待充电节点进行充电.实验结果表明,与传统Q学习方案相比,MC-CPP方案在获得了较优规划路径的同时,不仅能够减少充电无人机的数量和强化学习迭代次数,而且提高了无人机的能量利用率;与TSCA、NJNP、GC等方案相比,该方案能有效减少无人机飞行时间、节点死亡的数量及无人机能量消耗.

Aiming at the charging difficulties and low charging efficiency of wireless rechargeable sensor networks(WRSNs),a multi-UAV collaborative charging path planning scheme of wireless rechargeable sensor network is proposed.Firstly,the problem of multi-UAV path planning is described and a mathematical model is established.Then,the corresponding deep reinforcement learning(DRL)algorithm is proposed for the problem model.The algorithm uses neural networks,greedy strategies,and experience replay to obtain the flight path of the charging UAV.Finally,the system controls UAVs to charge the sensor nodes in the network along the planned path.The experimental results show that,compared with the traditional Q-learning scheme,the MC-CPP scheme can not only reduce the number of charging UAVs and the number of reinforcement learning iterations,but also improve the energy utilization of the UAVs while obtaining a better planning path.Compared with TSCA,NJNP,GC and other solutions,it can effectively reduce the flight time of UAVs,the number of dead nodes and the energy consumption of UAVs.