Research On UUV Cluster Task Planning Method Oriented To Recovery Task

In recent years,the technology of unmanned underwater vehicle cluster has developed rapidly and achieved remarkable results,which has become a major development direction of UUV technology.In the future,UUV clusters will inevitably become the main trend of future marine mission operations,and UUV technology will inevitably develop in the direction of distribution,clustering,autonomy,and miniaturization.And deployment and recycling in the flow of UUV’s mission is an essential part.Successful deployment and recovery are the basis for the normal operation of UUV.Therefore,in order to improve the utilization efficiency and guarantee capability of UUV clusters,this paper uses the recovery of UUV clusters as the research background to study the task planning system of UUV clusters.Fully consider the environmental adaptability and robustness of the UUV cluster to solve key issues such as task allocation,global planning,and local planning in an unknown environment.First,this paper establishes a UUV cluster task planning model based on recovery tasks.Establish a general task planning model based on the mixed integer linear programming method,and consider the dynamics and uncertainties in the UUV cluster recovery process,analyze the complexity of the task planning problem,and clarify the various constraints that the UUV cluster needs to meet during the recovery process.By analyzing the information coupling existing in task planning,a unified planning model for task allocation and trajectory planning based on the hierarchical decoupling idea and an integrated task planning model considering information coupling are established respectively.Then,when the UUV cluster is large,based on the hierarchical task planning model,the task planning process is hierarchically decoupled into two independent sub-modules:task allocation and trajectory planning.In terms of task allocation,a UUV cluster task allocation method based on a two-layer structure is proposed.The process of UUV cluster selection and recovery of the mothership is divided into the mothership distribution layer and the UUV cluster distribution layer.According to the different communication capabilities,the UUVs in the cluster are divided into two types:the "leader" who can communicate and interact with the recovery carrier and the "follower" who can only communicate with nearby UUVs.The mothership distribution layer is based on a centralized architecture,using an improved crow optimization algorithm(ICAS)to realize the pre-allocation of subgroups,and establish a oneto-one correspondence between UUV subgroups and the recovered mothership.The UUV cluster distribution layer is based on a distributed architecture.The "follower" UUV is defined as a task to be assigned,and the main constraint is the recovery capability of the mothership.The "Navigator" UUV uses a consensus binding algorithm based on team constraints(CBBAGC)to complete task assignment,and allocates a suitable recovery carrier for each UUV.In terms of trajectory planning,according to the amount of environmental information obtained,it is divided into global planning for the entire recovery task and partial planning for unknown threats.In the process of global trajectory planning,using the artificial potential field method as a tool,an improved algorithm based on "domain" is proposed.The method of calculating the intensity of the potential field is used to replace the traditional vector force control,and the speed potential field determined by the speed obstacle method is introduced.The proposed fish school trust region algorithm is used to calculate the sub-target points that meet the requirements,and the global trajectory planning of the UUV cluster in the known environment is completed.In order to deal with the unknown threats that appear during navigation,a rolling grid map environment modeling method based on quadtree is proposed.Use the quad-tree method to store scrolling environment information and reduce storage space.On this basis,an R-DRRT*algorithm is proposed to quickly optimize the local path.The IRRT*algorithm is used to make preliminary trajectory planning on the current raster map,and the elite set is obtained.The Gaussian mixture clustering algorithm is used to establish a sampling model based on learning.When an unknown threat appears in the grid map,the current search tree is pruned,and the DRRT*algorithm is used to quickly optimize the local path to realize the rapid avoidance of the threat.Finally,when the UUV cluster is small,by considering the coupling between task allocation and trajectory planning,the integrated method of UUV cluster task planning based on recovery tasks is studied.Establish an integrated mission planning overall framework by analyzing the recycling process.Use the improved vector histogram algorithm(VHF*)algorithm to build the track prediction matrix to increase the accuracy of the task assignment module.Considering the asynchronous communication between clusters,the asynchronous consistency bundling algorithm(ACBBA)algorithm is used to carry out the task allocation process,and the consistency process is transformed into a logic operation control network composed of multivariable fuzzy systems,and the logic dynamics is transformed by the semitensor product The system is transformed into an ordinary discrete dynamic system in nature.The verification module is added to the ACBBA algorithm to reduce the complexity of the algorithm by calculating the error value between the track estimation and the track planning.