贝叶斯网络结合决策理论的向前多步排故策略

针对序贯诊断、维修问题，提出基于贝叶斯网络和决策理论的向前多步排故策略生成算法.通过建立贝叶斯网络排故模型实现了不确定条件下排故知识的高效表达，同时使得推理算法与具体应用无关.采用决策影响图进行排故决策分析，充分利用观测操作间的相关性，选择合理的向前多步观测操作来降低维修盲目性.为了验证所提算法的有效性，采用随机排故策略、决策理论排故策略和理想排故策略的结果进行对比分析.仿真结果表明，所提算法通过增加合理观测操作，减少维修焦点和实际维修操作，使得总排故费用明显低于已有的启发式排故策略.

Bayesian networks and decision theory-based forward multi-step troubleshooting strategy

A forward multi-step troubleshooting strategy generation algorithm based on Bayesian networks and decision-theory was proposed for sequential diagnosis and maintenance problems. Troubleshooting knowledge under uncertainty was compactly represented by Bayesian network model and inference algorithm was independent on practical application. The correlation-ship among observations described in influence diagrams was explored to select reasonable forward multi-step observations and make troubleshooting decision in order to reduce blindness of repair. To verify the proposed method, the random troubleshooting strategy, decision theory strategy and ideal strategy were selected as comparison. Simulation results indicate that the proposed algorithm can significantly decrease the total troubleshooting costs by increasing the number of reasonable observer operation and reducing the numbers of maintenance focus and actual repair operation.