An Agent-Based Dynamic Scheduling Solution for Resource-Constrained Project Scheduling

Resource-constrained project scheduling (RCPSP) is aimed at achieving a valid scheduling solution to minimize the whole duration/cost of the project subject to resource constraint, precedence constraint and temporal constraint. As resource unavailability may be unknown in advance in some periods of the project in reality, an agent-based dynamic solution for RCPSP with task splitting and uncertain resource supply is provided to solve this problem, the essence of which is to dynamically adjusting the key tasks of the project in accordance with the tasks' time lags mainly caused by the uncertain resource supply. Besides, interruption of running tasks is allowed since it conforms to reality. A multi-agent system consisting of several agents and one resource agent is exploited to illustrate its performance. Lastly the experimental results illustrate the efficiency of dynamic solution.