基于元过程的多代理过程可靠性优化方法研究

Reliability Optimization Research on Multi-Agent Process Based on Meta-Process

为解决多代理过程可靠性难以优化的问题,提出了一种代理网络可靠性的闭环优化方法,其特点是引入了基于连通概率的可靠性计算方法.首先,通过目标输入启动可靠性闭环优化,将目标输入与系统可靠性参数经变换后进行处理,以作为新的给定值来控制多代理系统的网络调整.调整完成后,由可靠性计算获得新的系统数据,该数据再经变换、反馈后形成闭环控制系统.最终使得给定目标参数输入下的代理网络具有最佳可靠性.仿真结果显示,当活动代理数为100时,优化效果最为明显.

A closed loop optimization method of the reliability of agent network is proposed in order to solve the multi-agent process reliability problem which is difficult to be optimized. The feature of it is to introduce the reliability arithmetic based on connectedness probability. Firstly it initiates the reliability closed loop optimization with the target input, after handling the target input and system reliability through parameter transformation, the obtained value is regarded as a new given value to control the multi-agent network regulation, and the new reliability data are acquired by reliability calculation when the multi-agent network adjustment is completed. Then the data are retransformed and a closed-loop control system is formed through feedback. Finally, the agent network obtains the optimal reliability under the target input parameter. The simulation results show that system approaches to the best effect if the number of active agent is equal to 100.