A Study On The Grid Workflow Scheduling Modle Based On Particle Swarm Optimization Algorithm

Grid brings many changes to the workflow process, the maximal of which is the across-organization character of resources. So application of the traditional workflow technology in the gird environment must be faced to many problems, especially the Grid Workflow Scheduling, which affects the sucesseful excution and the efficency of the workflow. On the other hand, the execution of workflow activities needs to the persons'participation, but the persons are not able to continuously work without rest in 24 hours every day, 365 days every year. So, under the condition of the global collabaoration, we should consider many time factors caused by the resources arcoss-organization and across areas, such as the differences in time zone, time granularity and work time.Aiming at solving the above problems, the thesis studies grid workflow scheduling which consists of time sequencing and causality constraints, and build suitable object expressions for grid workflow scheduling , and analyse the object model from the differences in time zone, time granularity and work time, and furthermore find the most optimized exection path under the precondition of satisfying the object expressions . The main contents of the papere includes:(1) build an object model for grid workflow scheduling which consists of time sequencing and causality constraints, and solve the problem of outlining the uniform object expressions to the complex and multiple processes ; (2) discuss the object model from the several different aspects, including difference in time zone, time granularity, work time and holiday, and modify the object model from the above aspects; (3) According to the object model, apply the particle swarm algorithm to schedule grid workflow, then presents a modified particle swarm optimization algorithm, and design the optimization scheduling to the grid workflow by the modified particle swarm algorithm; (4) aiming at the two different grid workflow instances, does simulating experiments and analyze in matlab 7.0, and the proposed approach is verified and has the less execution time by comparing with other algorithm, the simulation results show the efficency of the PSO-based approach.