Model-based Diagnosis a failure is a discrepancy between the model and observations of an artifact a diagnosis restores consistency 1. Enumerate candidates in order of likelihood 2. Test novel failures by suspending constraints and testing consistency 3. Symptoms in candidate identify conflicting modes 4. Conflicts prune remaining candidates Consistency-based Diagnosis Ando Orl G(0 AndI Out(=In1( AND In2 U() .ALL components have “ unknown mode EOr .U never assigned in C Diagnosis =Al=G, A2=U O1=G, O2=U, 03=G Model u(X,Y over system variables X and mode variables y Obs AsSignment toOcX∪Y Candidate C:: Assignment of modes to Y Diagnosis D A candidate such that D1^Obs∧甲(X,Y) is satisfiable3 Model-based Diagnosis • A failure is a discrepancy between the model and observations of an artifact. • A diagnosis restores consistency. 1. Enumerate candidates in order of likelihood. 2. Test novel failures by suspending constraints and testing consistency. 3. Symptoms in candidate identify conflicting modes. 4. Conflicts prune remaining candidates. 4 Consistency-based Diagnosis And(i): ƒ G(i): Out(i) = In1(i) AND In2(i) ƒ U(i): ƒ Model: Ȍ(X,Y) over system variables X and mode variables Y ƒ Obs: Assignment to O Ž X ‰ Y ƒ Candidate Ci : Assignment of modes to Y ƒ Diagnosis Di : A candidate such that Di š Obs š Ȍ(X,Y) is satisfiable. 1 1 1 1 0 Or1 Or3 And1 A B C D E F G X Y Z 0 1 Diagnosis = {A1=G, A2=U O1=G, O2=U, O3=G} •ALL components have “unknown mode” U. •U never assigned in C