任务分配算法是多Agent研究的一个重要方向。基于传统合同网的任务分配算法在动态环境下存在效率较低的问题,而动态环境在实际工程中广泛存在。因此,为了使多Agent系统的任务分配算法适用于动态环境,提出了一种改进的合同网方法。该方法通过引入任务信任度和负载均衡度指标,对传统合同网的任务分配方法进行改进。仿真实验结果表明较现有方法具有更高的系统收益,且减少了多Agent系统任务完成所需时间。
Task-allocation is an important research direction of multi-agent system(MAS). Task-allocation algo-rithm based on traditional contract net protocol has low efficiency in dynamic environment. However, dynamic envi- ronment is popular in practice. To apply the task-allocation algorithm to dynamic environment, an improved con- tract net protocol is proposed by using the credit of task and the balance rate of load. The simulation results show that the method proposed does make the revenue increase and reduce the time needed for the task accomplishment.