Research On CPS Task Scheduling Algorithm Based On Improved Particle Swarm

The Cyber-Physical System(CPS)is a new type of distributed intelligent System that is produced under the background of social intelligent development.It takes information computing,real-time communication and intelligent control as the core to realize the fusion of information and physics.The reasonable task scheduling of CPS is the basis for the normal operation of the system.An efficient task scheduling strategy can improve the system's operating efficiency and stability,thereby improving system performance.However,The CPS has the characteristics of strong resource heterogeneity,high real-time perception and dynamic network topology,which makes the traditional task scheduling algorithm have the problems of low mapping efficiency and single demand constraint to dealing with CPS task scheduling.Based on this,this dissertation will study the CPS task scheduling problem from two aspects: establishing an efficient and reasonable mapping of CPS tasks and meeting the requirements of multiple Quality of Services(Qo S)of task users.The main research contents of this dissertation are:(1)First,we analyzed the basic structure,system characteristics and architecture about CPS.And we studied the task scheduling problem of CPS.Finally,we proposed an idea of applying Particle Swarm Optimization(PSO)to CPS task scheduling(2)In the process of CPS task scheduling,how to establish a quick and reasonable mapping between system resources and user tasks to reduce task time span.Aiming at this problem,an improved particle swarm task scheduling algorithm based on multiple swarm molecular dynamics theory was proposed.The algorithm introduces the molecular motion theory and optimization algorithm KMTOA into the particle swarm algorithm,First,the particle population is divided into different sub-populations,and then different gravitational rules are implemented in the sub-populations to guide the evolution of particles,which greatly improves the population diversity in the early stage and the local search ability in the later stage.Simulation and comparison experiments show that the improved scheduling algorithm has better task time span and load balancing ability,It can effectively improve the efficiency of resource mapping.(3)In the process of CPS task scheduling,due to the single task objective constraint,it cannot meet the user's multiple Quality of Service(Qo S).Aiming at this problem,an improved particle swarm task scheduling algorithm based on self-adaptive t-distribution is proposed.Firstly,the proposed algorithm introduces the mutation mechanism of self-adaptive t-distribution on the basis of elementary particle swarm,so as to improve the convergence speed and avoid the algorithm falling into local optimization.Secondly,when setting fitness function,the task scheduling is completed from three aspects: task completion time,task total cost and service quality.Finally,the task scheduling simulation experiment of the CPS is carried out,and compared with basic PSO algorithm and Cauchy mutation Particle Swarm Optimization(Cauchy-PSO)algorithm.Experimental results show that under the same experimental conditions,the proposed algorithm has better overall performance,and meets the user's multiple Qo S.