An Uav Inspection Task Allocation Mechanism Based On The Double-Layer Edge Networks

With the continuous promotion of energy Internet,the scale of power grid is gradually expanding,and the power department carries out power inspection tasks with the help of unmanned aerial vehicle(UAV)technology.However,with the continuous improvement of intelligence level,cloud computing technology cannot deal with the explosive growth of inspection task data,resulting in network congestion and large transmission delay.Mobile Edge computing Edge servers are deployed to provide close-range computing services for inspection tasks,reducing task transmission delays.Therefore,cloud computing technology and mobile edge computing should be comprehensively applied,and cloud-edge collaborative network should be adopted to provide fast and intensive computing services for inspection tasks.Cloud-side collaborative network needs a task allocation mechanism to reasonably allocate inspection tasks,but the existing task allocation mechanism has the problems of single optimization objective.and insufficient consideration of UAV mobility,which cannot meet the diversified demands of inspection tasks on delay and energy consumption.Aiming at the requirement of uav power inspection task for diversified delay and energy consumption,this paper proposes a uav inspection task unloading mechanism based on double-layer edge network.Firstly,the two-layer edge network model is designed,and the delay and energy consumption models are established.Then,using Lyapunov optimization theory,the long-term delay and energy consumption joint optimization problem is transformed into the Lyapunov drift penalty optimization problem of the current time slot.Finally,the unloading algorithm based on deep reinforcement learning algorithm is designed to solve the unloading decision.Experimental simulation results show that the proposed algorithm has lower average delay than other algorithms under the condition of meeting the energy consumption constraint,and the priority of delay and energy consumption is adjusted by weight parameters to meet the diversified demands of power inspection tasks.Furthermore,aiming at the problem of large processing delay of edge server caused by uav random mobility,this paper designed a uav inspection task migration mechanism based on edge collaboration.Firstly,uav mobility prediction model and edge server load model are established to predict uav movement direction and quantify the load of edge server.Then,the migration cost model of energy consumption and delay is established,and the migration decision problem is transformed into the migration cost minimization problem of the current slot edge server by using Lyapunov optimization theory.Finally,the deep reinforcement learning algorithm is used to solve the migration decision.Simulation results show that compared with other algorithms,the proposed algorithm has lower processing delay of edge servers and ensures load balance of edge servers.To sum up,this paper designs a task allocation mechanism for UAV electric power inspection,which has important theoretical application value,by comprehensively considering the diversified demands of UAV electric power inspection task on delay and energy consumption and the impact of UAV random mobility on task allocation.