New Methods for Deriving All Minimal Diagnostic Using Satisfiability Algorithms

In the model-based diagnosis reasoning, diagnosis in two steps,they are generating all minimal conflict sets of conflict identification and Generate all the minimal hitting sets of candidate generation.In this paper, we propose new method based on SAT solver generates all minimal diagnostic.Firstly the normal behavior,system model and obtained observations are described in conjunctive normal form.,then all related clauses of Pending diagnostic system put into SAT solvers. The decision circuit failure problem is converted to satisfiability problem. Hence combine CSSE-tree for solving minimal diagnostic.It can determine the point of failure without first solving the conflict set and then with hitting set algorithm.A method is presented to directly slove the minimum fault sets,which is quite different from the traditional model-based diagnosis.The diagnosis is only finished in one step.