【摘要】 将面向最经济服务的流演算机制引入到动态贝叶斯网络结构的学习中,提出一种面向最经济服务流的可视化动态贝叶斯网络分解协调模型(SFO-DBNs)及具体实现算法;该算法把Ford-Fulkerson流分解算法推广到多源、多汇的情况下,并加入了时间片t因素对服务流稳定性约束,可以把一个描述复杂大系统流演算的贝叶斯网络动态协调分解为几个子服务流分别建模,便于简化动态贝叶斯网络的构造与其推理机制.通过在数字气田采输管网调度中SFO-DBNs仿真实例分析,说明了该方法的有效性.更多还原

【Abstract】 Most economical service-oriented flows calculation mechanism is introduced during structure learning for dynamic Bayesian networks and a coordination decomposing model for service flows oriented dynamic bayesian network(SFO-DBNs) is presented.In this model,a complex flow-calculation for large system may be dynamically decomposed into several sub-service flows that can be respectively modeling,which extends algorithm Ford-Fulkerson to multi-sources and multi-sinks and joins the time slice t factors on the service flow stability constraint.Simulating for digital gas fields pipe-net emergent scheduling plan and pick-convey balance analysis illustrate that the method of SFO-DBNs is effective and efficient.更多还原