基于多智能体的公共检测资源协调方法

Multi-agent based coordination of public detection resources

公共检测资源协调过程的关键是任务合作者的选择和检测任务的分解.采用改进的合同网协议选择任务合作者,采用基于知识库的专家推理实现任务分解.在改进的合同网协调机制中,为每个任务类维护一个专家集,从而减少管理智能体的筛选负担和多智能体系统的通信开销,同时应用模糊综合评价方法实现投标书评价,并选取任务最佳合作者.采用多级推理机实现任务分解,设计了从实例中推导新知识和根据历史分解记录得到分解规则2种知识获取方法,设计采用正向推理策略的推理机.最后给出了公共检测资源协调的基于Web Service的实现.

The collaborator selection and task decomposition are important in the coordination process of public detection resources.The task collaborators were selected by using an improved contact net protocol,while the task decomposition was achieved by the expert knowledge based reasoning machine.In the improved contract net approaches,each task class has a set of experts whose values are defined by self-confidence,degree of trusted and degree of success,thereby reducing the burden of multi-agent system communication.In addition,the fuzzy comprehensive evaluation method was used to select the best task collaborators.In the task decomposition,the multi-level forward reasoning machine was used where new knowledge was obtained by deriving from the example and querying historical records.Finally,the realization of the coordination procedure was given by using Web service.