论文部分内容阅读
针对传统方法忽略测试通过的不确定性因素,缺乏长周期寻优机制,难以在复杂测试系统中生成全局最优诊断策略的问题,提出了一种基于马尔可夫决策过程(MDP)的诊断策略构建方法。该方法将故障检测、隔离的过程表述为系统故障状态的马尔可夫过程,通过引入折扣因子与目标权重,构造了综合效用准则函数的无限折扣模型,并利用策略迭代算法求解出全局平稳最优诊断策略。实例表明,该方法充分考虑了测试通过的不确定性,可实现全局平稳策略寻优,能够有效地指导测试系统实现快速故障检测和隔离。
Aiming at the problem that the traditional method ignores the uncertainty caused by the test, and lacks the long-period optimization mechanism, it is difficult to generate the global optimal diagnostic strategy in the complex test system. A diagnostic strategy based on Markov Decision Process (MDP) is proposed Build method. The method expresses the process of fault detection and isolation as the Markov process of system failure state. By introducing the discount factor and target weight, an infinite discount model of comprehensive utility criterion function is constructed and the global optimality Diagnostic strategy. The example shows that the method takes full account of the uncertainty of the test, which can achieve the optimization of the global smooth strategy and can effectively guide the test system to achieve rapid fault detection and isolation.