Research On Key Technologies Of Multi-agent Scheduling Constrained By Wireless Communications

Multi-agent systems is a group system composed of multiple autonomous agents that interact and cooperate with each other through wireless communication to complete complex tasks together.A multi-agent system integrating communication,control,artificial intelligence and other technologies can enhance the flexibility and reliability of task execution,ensure the stability and robustness of the system,and improve production efficiency and production income.Therefore,the multi-agent system has received great attention from the academic community,and has been widely used in the fields of smart factories,drone formations,and automatic driving.However,in order to achieve autonomous control,a multi-agent system needs to rely on wireless communication to achieve massive real-time data interaction.Therefore,the wireless communication in the system has high bandwidth and low delay requirements,and the characteristics of wireless communication often lead to communication interruption and communication congestion.The multi-agent itself also has problems such as system modeling,task assignment,policy control during scheduling.Therefore,the research on multi-agent scheduling technology under the condition of communication constraints has great significance and value.In this thesis,on the basis of extensive research on the characteristics and key technologies of multi-agent systems,aiming at the above problems and challenges,a task allocation algorithm and allocation strategy under different wireless communication multi-agent system models are designed,and the system scheduling and communication problems are solved to ensure the optimal control performance and high robustness of the system.First,in order to analyze the system model of the smart factory and solve the problem of task allocation,this thesis designs a heterogeneous multi-agent system model,and establishes a mapping matrix between the task function requirements and the characteristics of the agents in the system.The control constraints are obtained by encoding the system tasks,and a task assignment algorithm and a task execution algorithm based on control optimization problems in heterogeneous multi-agent systems are proposed.At the same time,the random model of agents and tasks in the system is designed,and the average control cost of the system is analyzed statistically.The simulation results prove the reliability and superiority of the allocation algorithm and stochastic model.Secondly,because terahertz has the advantages of ultra-high bandwidth and ultralow delay,which can meet the demand of massive real-time data transmission in smart factories,this thesis designed a multi-agent system model based on terahertz communication,and proposed a task allocation algorithm based on weight under the condition of comprehensive consideration of control cost and communication cost.This algorithm can solve the task assignment problem in the terahertz communication model and reduce the influence of communication congestion to a certain extent.In order to further solve the problem of communication congestion and dynamic interference in the system,this thesis designs an elastic allocation strategy based on error thresholds by quantifying the performance errors caused by congestion and interference.The simulation results show that the weight-based task allocation algorithm is obviously superior to the traditional control cost-based allocation algorithm in terms of comprehensive control cost and task completion.At the same time,the flexible allocation strategy can improve the task completion of the system and ensure the robustness of the system.