Research Of Multi-Agent Task Allocation For Process Management System

Task allocation is an important issue of modern intelligent management system,especially in process management system. Conducting the efficiency of task allocationprocess, exerting the abilities of different machines, and optimizing the environmentwith resources, are important parameters to measure the level of working efficiency. Onthe basis of Multi-Agent technology, this paper makes a study on the formal modellingand model resolving in this process management system.Firstly, the research of task decomposition based on Multi-Agent theory is thepreparatory work of task allocation. The original complex tasks are divided into simpletasks which one Agent can execute independently; changing the complex task allocationproblem into simple task allocation is the main purpose of task decomposition.Then, by combining contract net protocol models with the main function andarchitecture of the process management system, promoted method of task allocationbased on contract net and distributed Agent collaborative network were proposed, inorder to solve the task allocation problems in intelligent systems.On the basis of the contract net protocol model, multi-Agent task allocation based ontime and resource was proposed. By generalizing consumptions in the process of taskallocation as time consuming and resource consuming, this paper build up amulti-Agent task allocation model, and verified its correctness and efficiency. Then,improved genetic algorithm was designed to solve this task allocation model.In order to realize the application of this multi-Agent task allocation model, JADEsoftware was used as Agent platform. Secondary development of this processmanagement system was posed with an application example. Through this example, theefficiency of task decomposition and task allocation model was verified.