A Hybrid Multiagent-Based Rescheduling Mechanism for Open and Stochastic Environments Concerning the Execution Stage
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Dynamic scheduling is a hot topic in recent years. Most existing dynamic scheduling algorithms process the rescheduling in an off-line way through the modification of created schedules in advance of further execution. However, in open and stochastic environments, such as manufacturing systems, streams of exogenous events may cause information changes inside of jobs and resources during the job execution stage, so as to increase the environmental uncertainties and rescheduling computational complexity. In such environments, the frequent occurrence of dynamic events requires scheduling methods to provide efficient and robust online rescheduling services for ensuring the success rate of jobs. This paper investigates the dynamic events during the execution stage in open and stochastic environments and proposes a multiagent-based rescheduling mechanism with the hybrid structure for such cases. Three types of agents are proposed to implement the rescheduling mechanism. The experimental results demonstrate that the proposed hybrid mechanism own higher efficiency and robustness in dealing with dynamic events in open and stochastic scheduling environments than that of existing centralized and decentralized methods.
Open Access Status
This publication is not available as open access