The Research Of Manufacturing Grid Scheduling Algorithm Based On Workflow

The strategy of "Information Technology Gives an Impetus to Industrialization and Industrialization Promotes Information Technology" proposed in our country enables computer technology more and more widely used in modern manufacturing. Traditional manufacturing has characteristics of information silo between enterprises, irrational use of manufacturing resources, higher manufacturing cost and confusion of the management of manufacturing processes. As the development of grid and workflow technology, the introduction of grid workflow technology in manufacturing is significant. It enables enterprises share information, use the global resources transparently, effectively and automatically manager the manufacturing workflow, in order to reach the goal of cutting cost and improving efficiency.In the background of economic globalization, different service types, different level of quality of service (QoS) of manufacturing resources distribute in various regions all over the country and all sectors, forming the tremendous manufacturing grid resources. How to realize automatic management of workflow, tasks scheduling and grid resources allocation with the grid workflow technology, and further create a optimal production plan which meets the requirement of cutting cost, improving the level of QoS and rational use of resources, is highly discussed in our paper.The research of our paper mainly focuses on manufacturing workflow modeling, grid workflow scheduling algorithm and the grid workflow schedule system prototype. The main work and creation are as following:1) Utilize DAG to model the processes of manufacturing tasks, and add deadline constrain property to DAG.2) Propose a QoS-constrained workflow scheduling algorithm—TP(Task Partition) algorithm. TP meets the deadline of the workflow while paying less than the current scheduling algorithm.3) Construct grid workflow schedule system prototype. The system implements basic functions, TP and D-MDP algorithm.4) Combine the grid workflow schedule system with grid simulation platform GridSim to test our TP algorithm. The experiment results show that the improved TP algorithm costs less than D-MDP, thus, have better performance on cost.