Reliability Optimization of Safety Monitoring and Protective Systems

本研究提出一种安全监控系统的最优逻辑结构，以使故障危险故障或故障安全故障所造成的预期总损失最小化。考虑了由多个通道组成的安全监测系统。问题是在各通道按照事故发生机制逻辑连接的情况下，如何设计各通道的最优连贯结构。对于单通道系统，证明了最优通道结构为<k^*> -out-of-n: G，且<k^*>可解析计算。对于多信道系统，最优信道结构为k-out (n: G)，并将其表述为非线性整数规划问题，可以用扩展的Lawler和Bell方法求解。上述结果被推广到涵盖更一般的逻辑结构优化，而不假设自组织通道连接。为整个保护系统开发了最佳的关闭逻辑，该系统由驱动、判断和传感部分组成。最佳关断逻辑由开关函数决定。对于具有可靠判断段和可靠驱动段的系统，可以根据感知段的可靠性信息确定最优停机逻辑。为安全监测系统开发了最优动态逻辑，其中的“动态”是指每次获得新的被监测数据时都要重构逻辑。最优动态逻辑是由监测数据历史的切换函数决定的，并且在下一次监测时，它使两个故障造成的预期损害最小化。为了提高设备的可用性和安全性，提出了多传感器系统诊断的最优决策逻辑。由两种故障模式引起的预期损失被最小化，并且由此产生的最优诊断逻辑由切换函数的特征决定，该开关函数反过来表示诊断的效率。如果利用测量数据的重要性排序，则可以大大简化最优逻辑的解析表示。少

This study introduces an optimal logic structure of a safety monitoring system to minimize the expected total loss incurred upon fail-dangerous failures or fail-safe failures.1. The safety monitoring system composed of several channels are considered. The problem is to design an optimal coherent structure for each channel when the channels are logically connected according to an accident occurence mechanism. For the one-channel system, the optimal channel structute is proven to be <k^*> -out-of-n: G, and <k^*> can be calculated analytically. For the multi-channel system, the optimal channel structure is a k-out-of-n: G, and the problem is formulated as a non-linear integer programming which can be solved by the extended Lawler and Bell's method. The above results are generalized to cover more general optimization of the logic structure without assuming ad-hoc channel connections.2. The optimal shut-down logic is developed for the overall protective system, which is composed of the driv … More ing, judging, and sensing sections. The optimal shut-down logic is determined by a switching function. For the system with reliable judging and driving sections, the optimal shut-down logic can be determined based on the reliability information of the sensing section.3. The optimal dynamic logic is developed for safety monitoring systems, where the term "dynamic" means that the logic is reconstructed every time when new nomitored data are obtained. The optimal dynamic logic is determined by a switching function of a history of monitored data, and it minimizes an expected damage caused by the two failures at the next monitoring time.4. The optimal decision logic for diagnosis of multi-sensor systems is developed to improve plant availability and safety. An expected loss caused by the two failure modes is minimized, and the resultant optimal diagnostic logic is determined by signatures of a switching function which, in turn, expresses efficiency of the diagnosis. An analytic representation of the optimal logic can be simplified considerably if importance rankings of measurement data are exploited. Less