Research And Implementation On Correlation Dynamical Diagnosis Of Discrete-Event Systems

Intelligent diagnostic problem is one of the most important application problems inArtificial Intelligence. With the improving requirements for the security and reliabilityof dynamical system, more and more attentions are paid to the methods of thedynamical diagnosis in the discrete-event systems. Discrete-event systems are indiscrete state and event-driven. The diagnosis for the kind of systems is calleddynamical diagnosis.The paper proposes the methods of correlation dynamical diagnosis. This methodidentifies the associated failure events based on the observation, exclude the impossiblefault, and then enumerate the corresponding fault set. It uses the pruning rules to get thefailure sequence assumptions. And finally verify whether the failure sequenceassumption is consistent with the observation of the system.The paper particularly focuses on the following aspects of the dynamical diagnosis.First, give the formal definitions and models of correlation dynamical diagnosis. Second,demonstrate the relation of the associated set of fault events and minimal diagnosis, andthen explain the solution of minimal diagnosis with correlation dynamical diagnosis.Third, propose the key algorithm of correlation dynamical diagnosis, the algorithm toget all diagnosis and minimal diagnosis of the system. In addition, analysis thecomplexity of the algorithm mentioned above. In the end of this paper, realize thecorrelation dynamical diagnosis algorithm, and solve the diagnosis problems of thesystem. In the experiment, this paper gives useful pruning rules to reduce theenumeration space. By comparing dynamical diagnosis algorithm based on satisfiability,the experiment’s result shows the correlation dynamical diagnosis algorithm is moreeffective in solving diagnosis problems.