Research On Workflow Scheduling Method For Industrial Cloud Environment

A workflow is a model consisting of a series of tasks to be executed in a certain order,which is used to describe the execution process and each operation step.With the development of cloud computing technology and its application in industry field,industrial cloud platforms can meet personalized computing requirements and provide computing power services for computing tasks in industrial manufacturing,with more and more workflows are offloaded to the cloud platform for execution.The workflows under the industrial cloud are characterized by complex structures and large computational scales,but the traditional scheduling algorithms have the disadvantages of easily falling into the local optimal solution.In addition,most workflow scheduling research still focus on single-objective optimization of execution time,ignoring other performance metrics,resulting in high cost and low resource utilization caused by the single pursuit of timeliness.Based on the above problems,this thesis conducts research on workflow scheduling under industrial cloud,and the main work is as follows:(1)In response to the single problem of the workflow scheduling target under the industrial cloud,this thesis builds a multi-objective trade-off scheduling model for industrial cloud based on the traditional scheduling model,considering three optimization objectives of workflow execution time,execution cost and load balancing at the same time,and designing a nonlinear decreasing weight particle swarm algorithm with multiple swarm collaboration(MSPSO-ND)to solve it.The algorithm is improved by chaotic initialization,master-slave population strategy and nonlinear decreasing inertia weights.The experimental results show that the algorithm has better results in reducing scheduling execution time and execution cost than other scheduling algorithms.(2)In response to the problem of low resource utilization rates in the workflow scheduling,this thesis constructs a multi-objective trade-off workflow scheduling model for resource-constrained scenarios,with scheduling execution time,execution cost and resource utilization as optimization objectives,and proposes an improved gray wolf optimization algorithm(LPSO-GWO)based on particle swarm to solve it.This algorithm provides a high-quality initial solution for the gray wolf by a particle swarm algorithm.The inertia constant is added in the gray wolf iteration,and the gray wolf step size is adjusted by incorporating the random wandering strategy to ensure the global search capability and local exploitation capability of the algorithm in the later stage.The experimental results show that the algorithm can effectively improve the resource utilization.(3)This thesis designs and implements a workflow scheduling verification system for industrial clouds.The workflow scheduling module is designed and implemented based on the two scheduling algorithms proposed in this thesis,and the system is developed using the front-and back-end separation mode.The feasibility of this system for workflow task scheduling is verified through the implementation and demonstration of relevant modules.