Efficient Mincuts Identification for Phased-Mission Systems

Yuchang Mo,Jinping Liao,Xuli Liu,Liudong Xing

doi:10.1109/access.2020.3045283

Yuchang Mo, Jinping Liao + Show 2 more

Open Access

https://doi.org/10.1109/access.2020.3045283

Copy DOI

Abstract

Fault Tree is an important model for reliability and safety assessment. The analysis performed on a fault tree can be either qualitative or quantitative. Both types of analyses may involve identifying minimal cut sets (MCS) or mincuts, each of which is a minimal combination of basic events (component failures) whose occurrence causes the top event (system failure) to occur. Considerable research efforts have been expended in the identification of MCSs for single-phased systems and networks. However, only little work is available for phased-mission systems (PMSs) and the existing method involves a large number of redundant computations in the MCS identification for eliminating redundancies across generated cut sets. This article proposes an MCS identification method based on binary decision diagrams (BDD) generated from the PMS fault tree using the backward ordering. By examining cut sets encoded by the generated BDD model, we identify two kinds of redundancies (inclusion relation-based and intra-implication relation-based) that prevent a cut set from being an MCS. Correspondingly, two BDD operations are developed for eliminating these two kinds of redundancies, contributing to correct and efficient MCS identification. As demonstrated through experiments on a highly-reliable distributed computing infrastructure PMS, the proposed MCS identification method is more efficient than the existing method.

Highlights

A phased-mission system (PMS) is a system subject to multiple, consecutive, non-overlapping phases of operation [1, 2]
We propose an minimal cut sets (MCS) identification method based on a binary decision diagrams (BDD) generated from the PMS fault tree using the backward ordering [16]
Based on the examination of the cut sets encoded by the generated PMS BDD model, we identify two kinds of redundancies that can prevent a cut set from being an MCS

Summary

INTRODUCTION

A phased-mission system (PMS) is a system subject to multiple, consecutive, non-overlapping phases of operation [1, 2]. Recent work [15] proposed to identify MCSs for a PMS based on the PMS BDD model generated from the PMS fault tree using the forward ordering heuristics (the order of variables of the same component is the same as the phase order) This method often generates a large PMS BDD model, based on which the MCS identification involves a large number of redundant computations for eliminating interimplication relation-based redundancies (such redundancies caused by the forward dependencies of component states across the phases will prevent a cut set from being an MCS). Two BDD operations, named IncRed and InImpRed in this work, are developed for eliminating these two kinds of redundancies They are combined to identify MCSs correctly from a PMS BDD model generated using the backward ordering.

PMS and PMS Fault Tree

REDUNDENCY IDENTIFICATION AND BDD OPERATIONS

MCS IDENTIFICATION

CASE STUDY

MCS IDENTIFICATION The following steps are conducted to identify all MCSs

CONCLUSION