| As one of the most influential high technologies,wireless sensor networks(WSNs)have changed the world in the 21 st century.As the second largest network after the Internet,WSNs have attracted worldwide attention and laid the foundation for the vigorous development of the Internet of Things,big data and next-generation artificial intelligence.Although the application prospect of WSNs is very promising,the energy consumption and maintenance of sensors is still a bottleneck affecting their large-scale deployment and widespread use.The rapid development of wireless charging technology and unmanned aerial vehicles(UAV)provides a new type of energy supplement method for all kinds of WSNs,which can greatly improve the quality of service,extend the service lifetime and improve the stable performance of WSNs.Wireless charging devices carried by UAVs can assist WSNs to achieve multidimensional aerial charging.But for practical application requirements,how to plan the charging path of UAV is a key technical problem.Focusing on the problems mentioned above,this dissertation focuses on the relevant models and algorithms of UAV aided charging planning for WSNs.The UAV assisted WSNs permanent cycle charging planning problem,the one-to-many coverage charging planning model and algorithm for UAV assisted intensive WSNs,charging planning of sustainable and heterogeneous UAV with multiservice station for sparse WSNs,and the random dynamic multi-UAV path planning problem are comprehensively studied in turn in the dissertation.The proposed models and algorithms have certain universality,and can also be applied to other optimization problems with the similar model characteristics by making slightly modifications.The main content and innovation points in the dissertation are as follows:(1)UAV-aided WSNs permanent cycle charging planning algorithmIn order to meet the lifetime requirements of WSNs,the definition of UAV-aided WSN permanent cyclic charging planning for WSNs is given,the sufficient necessary conditions for an UAV-assisted WSN permanent cyclic charging planning is proved,and a UAV-assisted WSNs permanent cyclic charging planning decision algorithm and implementation method are proposed to minimize the total working time of UAV.The experimental results indicate that,with a large ratio of charging time to flight time,the improvement of UAV charging efficiency should focus on reducing charging time rather than optimizing flight path,which also lays the foundation for the follow-up research of UAV-assisted service in WSNs.(2)Hierarchical joint optimization algorithm for UAV one-to-many coverage charging planningAiming at the efficient charging requirements of dense WSNs,a one-to-many coverage charging planning model of UAV is established,and a hierarchical joint optimization algorithm framework are designed,which is constructed by sensor node clustering algorithm,segmented adaptive firefly algorithm and UAV hovering position adjustment algorithm.A clustering method is adopted to cluster the sensor nodes for one-to-many charging.A segmented adaptive firefly algorithm is proposed to solve the UAV charging path and an algorithm to adjust the hovering position of UAV is proposed to improve the charging rate and reduce the charging time in the cluster.The experimental results show that the charging efficiency is improved and the energy loss in the charging process is reduced by one-to-many covering charging strategy and the optimal UAV hovering position.(3)Time first charging planning method for sustainable heterogeneous UAVs with multiple service stationsFor the service requirements of sparse WSNs,the heterogeneous UAV system with multiple service stations is considered.UAVs are allowed to charge themselves nearby during the task.A sustainable heterogeneous UAV charging planning model with multiple service stations is established with the goal of completing all tasks as soon as possible.A time series first immune clonal selection algorithm with optimization and modification is designed.The time first partitioning sequence algorithm can allocate sensor nodes more evenly for heterogeneous UAV.The optimization correction algorithm adjusts the flight path of each UAV through a variety of variational operators to help the immune clone selection algorithm jump out of the local optimum.The comparison and analysis of experimental results verified the good optimization ability of the algorithm.(4)Cooperative decision model for multi UAV stochastic dynamic charging planningAiming at the multi-UAV stochastic dynamic charging planning problem,combined with the efficient cooperation ability of multi-agent and the rapid real-time decision-making ability of deep reinforcement learning,a multi-UAVs cooperative decision model based on Markov game and transformer is designed.An experience extraction A3C(Asynchronous Advantage Actor-Critic)algorithm is proposed,in which an experience pool is introduced into the algorithm.With the evolution of the algorithm,the ability of experience pool to guide training model parameters is gradually reduced.The simulation results indicated that the speed of the proposed model is much faster than those of other intelligent optimization methods such as Google OR Tools,LKH3 and so on. |
