Research On Imperfect Fault Coverage Models Considering Failure Propagation Effects

With the continuous improvement of modern industrial level,fault-tolerant technology has been widely used in many large-scale task systems,fault detection,location,isolation and reconfiguration are important means to achieve system fault tolerance.The model that takes the imperfectness of fault coverage into account is called the Imperfect Fault Coverage Model(IFCM).In IFCM,the failure of an uncovered component will directly lead to the entire system failure,which is called uncovered failure.This is because if a component failure is propagated,the failure of the component,if not successfully isolated,will harm other normal components and lead to system failure,such as un-isolated fires,leaks,or viruses;furthermore,if the reconfiguration fails,for example a failure is undetected and cannot be reconfigured,even if the system has sufficient redundant spare parts to be used,the system will fail because the spare parts cannot replace the failed components to work.IFCM provides a theoretical basis for system reliability analysis under imperfect fault coverage,but it does not further analyze the relationship between component failure propagation and component relevance,fault coverage processes,and functional dependencies,and their impact on system unreliability.Specifically,IFCM does not consider the relevance changes of components,and ignores the failure of irrelevant components.If the failure of irrelevant components is not propagated,even if the coverage fails,it will not affect the reliability of the system;in addition,IFCM assumes that isolation or reconfiguration failure will lead to non-covered failure,does not consider the fault coverage process of different isolation and reconfiguration sequences,there may be cases where reconfiguration fails after successful isolation and failure of isolation after successful reconfiguration will not lead to the system failure.For example,irrelevant components only need to be isolated successfully,even if the reconfiguration fails,it will not affect the system reliability because of its irrelevance in the system;in addition,the existing IFCM-based functional dependency analysis assumes that the failure of the triggering component(functionally depended component)will lead to the inaccessibility or unavailability of its dependent components,so that the dependent component can no longer have uncovered failures,but this may overestimate the reliability of the system,because being disconnected by a function(path)in a functional dependency is not exactly equivalent to the safe isolation of failed components by fault-tolerant mechanisms.In view of the above problems,this thesis further studies the imperfect fault coverage model considering component failure propagation.The specific work is as follows:1)Aiming at the problem that IFCM does not consider component relevance and component failure propagation,it is proposed to divide component failures into propagating failures and non-propagating failures,and an imperfect coverage model that considers component relevance and failure propagation is proposed.Compared with the traditional IFCM,in the model of this thesis,if the failure of irrelevant components is not propagated,it will not lead to the system failure even if it is not covered,so that more accurate system reliability analysis results can be obtained.2)Aiming at the different fault coverage processes,combined with the propagation of component failure and the relevance of components,two different imperfect fault coverage models of first isolation and then reconfiguration and first reconfiguration and then isolation are analyzed,and a more accurate system reliability analysis method is given.Compared with the traditional IFCM,in the imperfect coverage model of first isolation and then reconfiguration proposed in this thesis,the failure of irrelevant components only need to be isolated successfully without causing system failure;in the imperfect coverage model of first reconfiguration and then isolation proposed in this thesis,if non-propagating component failure is reconfigured successfully,it will not cause the system to fail even if it is not isolated.3)For functional dependency,it is pointed out that functional disconnection in functional dependency is not totally equivalent to safety isolation in fault coverage mechanism,and an imperfect fault coverage model considering functional dependency and failure propagation is proposed.Compared with traditional IFCM,in the model of this thesis,functional disconnection generally only causes the normal function of(functional)dependent components to be inaccessible,and cannot safely isolate the propagating failure of dependent components,such as fire or leakage,because these propagating failures do not necessarily propagate along functionally dependent pathways.The case analyses show that the imperfect coverage model proposed in this thesis can calculate the more accurate system reliability due to considering the influence of the uncovered failure of dependent components on the system unreliability under the condition of functional disconnection.