Font Size: a A A

Multi-sensor Resource Scheduling Method Research For Cooperative Detection Of Flying Targets

Posted on:2022-06-21Degree:MasterType:Thesis
Country:ChinaCandidate:M Q WangFull Text:PDF
GTID:2492306536967419Subject:Engineering (Control Engineering)
Abstract/Summary:PDF Full Text Request
The continuous development of aircraft technology targeting low altitude,stealth and high speed has brought huge challenges to the existing detection and surveillance sensor systems.Therefore,the current low-altitude sensor detection equipment has launched a huge challenge.According to the traditional resource scheduling project of independent detection by a single detection platform in the past,it has been unable to solve the current flying target detection problem.In addition,due to the maneuverability of the flying target,the demand for real-time sensor resource scheduling is increasing,resulting that existing countermeasures against flying targets are overwhelming.This article is mainly based on the background of multi-sensor cooperative detection of flying targets,guided by the problems in the process of multi-sensor cooperative detection resource scheduling,researching multi-sensor cooperative detection resource scheduling method and system construction in order to provide new solutions for resource scheduling problems.The main results of the research are as follows:(1)In research of the problem of multi-sensor cooperative resource scheduling.The coverage detection indicator is introduced,the influencing factors of multi-sensor cooperative resource scheduling are summarized,and the multi-sensor cooperative resource scheduling index model is constructed.In order to realize the omni-directional coverage detection of the aircraft,the coverage detection index is constructed based on the multi-sensor cooperative detection technology,and the constraint conditions affecting the resource scheduling process are sorted out,and the evaluation index is formed.Finally,based on the tomographic analysis method,a multi-sensor cooperative detection resource scheduling evaluation index model is constructed,which lays a solid foundation for the construction of the dynamic scheduling process model.(2)In research of the modeling method of multi-sensor cooperative detection resource dynamic scheduling process.Firstly,the basic principles of Markov Decision Process(MDP)for modeling the dynamic scheduling process of multi-sensor cooperative detection resources are analyzed,and the basic elements of MDP modeling are described.Secondly,combined with the background of multi-sensor cooperative detection resource scheduling for flight targets and the basic elements of MDP modeling,relevant state variables are defined,and the related evaluation indexes are abstracted into mathematical models.Combined with multi-sensor cooperative detection resource scheduling evaluation index model,the reward function for evaluating the benefit of the dynamic scheduling process of multi-sensor cooperative detection resources is formed,and the dynamic scheduling process model of multi-sensor cooperative detection resources oriented to flying target detection is constructed.Corresponding standards and cornerstones are provided for subsequent algorithm design,algorithm verification,and algorithm performance comparison.(3)In research of the resource scheduling method of multi-sensor cooperative detection.Firstly,the basic principles and classification of reinforcement learning based on Neural Network are analyzed.Secondly,in order to solve the problem of poor real-time performance in resource scheduling,the multi-sensor cooperative detection resource scheduling algorithm based on PPO-FCNN(Proximal Policy OptimizationFully Connected Neural Network,PPO-FCNN)is proposed.The decision-making network training method and algorithm training architecture are optimized to improve the stability of the algorithm and form a stable resource scheduling solution algorithm.Finally,through comparison with the simulation results of existing resource scheduling algorithms(such as genetic algorithm,DQN(Deep Q Learning,DQN)),the validity and stability of the multi-sensor cooperative detection resources dynamic scheduling process model and the algorithm are verified.(4)In research of the multi-sensor cooperative detection resource scheduling prototype system design.Firstly,the prototype system design method is researched and proposed,including the layer of development platform,the layer of database,the layer of basic function,the layer of system function.Secondly,with the layer of system function as the leading content,the design concepts and ideas of the four major function modules of the prototype system are described.Finally,in the multi-sensor cooperative detection resource scheduling simulation environment,limited sensor resources are realized usefully,and the management and monitoring of target and detection equipment status information is centralized.
Keywords/Search Tags:Multi-sensor collaboration, Reinforcement learning, Resource scheduleing, Markov decision process, Adam optimization
PDF Full Text Request
Related items