Model based Diagnosis The Single Fault Case Brian C Williams 16 412J 6 834J October 9th 2002 9 13 00 courtesy of JPL copyright Brian Williams 2000 1 Brian C Williams copyright 2000 WORLD observations Plant sense Diagnostic Agent Probes and Tests 9 13 00 act AGENT Monitors Diagnoses Repairs Avoids P s actions Symptom based Consistency based copyright Brian Williams 2000 2 Outline Single Fault Diagnosis Explaining the symptoms Handling the unknown 9 13 00 copyright Brian Williams 2000 3 Failures may be hidden STS 93 Symptoms Engine temp sensor high LOX level low GN C detects low thrust H2 level low Problem Liquid hydrogen leak Effect LH2 used to cool engine Engine runs hot Consumes more LOX 9 13 00 copyright Brian Williams 2000 4 What is Fault Diagnosis 3 A 2 B 2 C 3 D 3 E 9 13 00 M1 X A1 M2 Y A2 M3 F 10 G 12 Z copyright Brian Williams 2000 5 What is Fault Diagnosis Model based Diagnosis Given a system with symptomatic behavior and a model of the system find diagnoses that account for symptoms 3 A 2 B 2 C 3 D 3 E 9 13 00 M1 M2 6 6 X A1 Y A2 M3 12 F 10 Symptom G 12 Z copyright Brian Williams 2000 6 What is Fault Diagnosis Model based Diagnosis Given a system with symptomatic behavior and a model of the system find diagnoses that account for symptoms 3 A 2 B 2 C 3 D 3 E 9 13 00 M1 M2 6 6 X A1 Y A2 M3 12 F 10 Symptom G 12 Z copyright Brian Williams 2000 7 Standard Diagnostic Approach 1 Generate candidates from symptoms 2 Test if candidates account for symptoms Diagnoses should be complete Diagnoses should exploit all available information 9 13 00 copyright Brian Williams 2000 8 Outline Single Fault Diagnosis Explaining the symptoms Handling the unknown 9 13 00 copyright Brian Williams 2000 9 How Should Diagnoses Account for Symptoms Abductive Diagnosis Given symptoms find diagnoses that predict symptoms 3 A 2 B 2 C 3 D 3 E M1 M2 6 6 X A1 Y A2 M3 12 F 10 Symptom G 12 Z M1 Drops low bit on A 9 13 00 copyright Brian Williams 2000 10 Single Fault Abductive Diagnosis by Generate and Test Given Exhaustive fault models for each component Generate Consider each single fault mode as a candidate Test 1 Simulate candidate given inputs 2 Compare to observations Disagree Discard Agree Keep No prediction Discard If all fault models discarded the component okay exonerate Problem Fault models may be incomplete May incorrectly exonerate faulty components 9 13 00 copyright Brian Williams 2000 11 Outline Single Fault Diagnosis Explaining the symptoms Handling the unknown 9 13 00 copyright Brian Williams 2000 12 Failures are Often Novel Mars Observer Mars Climate Orbiter Mars Polar Lander Deep Space 2 courtesy of JPL 9 13 00 copyright Brian Williams 2000 13 9 13 00 copyright Brian Williams 2000 14 How Should Diagnoses Account for Novel Symptoms Consistency based Diagnosis Given symptoms find diagnoses that are consistent with symptoms Constraint Suspension Make no presumptions about faulty component behavior 3 A 2 B 2 C 3 D 3 E 9 13 00 M1 M2 6 6 X A1 Y A2 M3 12 F 10 Symptom G 12 Z copyright Brian Williams 2000 15 How Should Diagnoses Account for Novel Symptoms Consistency based Diagnosis Given symptoms find diagnoses that are consistent with symptoms Constraint Suspension Make no presumptions about faulty component behavior 3 A 2 B 2 C 3 D 3 E 9 13 00 M1 M2 6 6 X A1 Y A2 M3 12 F 10 Symptom G 12 Z copyright Brian Williams 2000 16 How Should Diagnoses Account for Novel Symptoms Consistency based Diagnosis Given symptoms find diagnoses that are consistent with symptoms Constraint Suspension Make no presumptions about faulty component behavior 3 A 2 B 2 C 3 D 3 E 9 13 00 M1 M2 6 6 X A1 Y A2 M3 F 10 G 12 Z copyright Brian Williams 2000 17 Single Fault Diagnosis as Constraint Suspension 1 Find symptoms Simulate forward limited inference 2 Generate candidates from symptom Identify components in conflict 3 Test candidates against all observations 1 Suspend model of candidate 9 13 00 copyright Brian Williams 2000 18 1 How Do We Find Symptoms Simplest Forward simulation from inputs 3 A 2 B 2 C 3 D E 9 13 00 6 M1 12 X 6 M2 A1 Y A2 M3 6 Z copyright Brian Williams 2000 F 10 G 12 12 19 1 How Do We Find Symptoms Simplest Forward simulation from inputs Compare predictions to outputs 3 A 2 B 2 C 3 D E 9 13 00 6 M1 12 X 6 M2 A1 Y A2 M3 6 Z copyright Brian Williams 2000 F 10 G 12 12 20 2 Generate Candidates Identify Conflicts 3 A 2 B 2 C 3 D E 6 M1 X 6 M2 12 A1 Y F 10 G Z Symptom F is observed 10 but should be 12 if M1 M2 A1 are okay Conflict 9 13 00 ok A1 ok M1 ok M2 is inconsistent copyright Brian Williams 2000 21 2 Generate Candidates Identify Conflicts 3 A 2 B 2 C 3 D E 6 M1 X 6 M2 12 A1 Y F 10 G Z Symptom F is observed 10 but should be 12 Conflict ok A1 ok M1 ok M2 is inconsistent Candidates not ok A1 or not ok M1 or not ok M2 9 13 00 copyright Brian Williams 2000 22 3 Test Candidates Constraint Suspension 3 A 2 B 2 C 3 D 3 E M1 M2 M3 X A1 F 10 12 A2 G 12 12 Y Z Select candidate M1 9 13 00 copyright Brian Williams 2000 23 3 Test Candidates Constraint Suspension 3 A 2 B 2 C 3 D 3 E X A1 M2 Y A2 M3 F 10 G 12 Z Select candidate M1 Suspend M1 s constraints 9 13 00 copyright Brian Williams 2000 24 3 Test Candidates Constraint Suspension 3 A 2 B 2 C 3 D 3 E X M2 M3 6 6 A1 Y A2 Z F 10 G 12 12 Select candidate M1 Suspend M1 s constraints Predict forward 9 13 00 copyright Brian Williams 2000 25 3 Test Candidates Constraint Suspension 3 A 2 B 2 C 3 D 3 E M1 X M2 Y F 10 A2 M3 G 12 Z Select candidate A1 Suspend A1 s constraints 9 13 00 copyright Brian Williams 2000 26 3 Test Candidates Constraint Suspension 3 A 2 B 2 C 3 D 3 E 6 M1 M2 M3 6 6 X F 10 Y A2 Z G 12 12 Select candidate A1 Suspend A1 s constraints Predict forward 9 13 00 copyright Brian Williams 2000 27 3 Test Candidates Constraint Suspension 3 A 2 B 2 C 3 D 3 E M1 X A1 Y A2 M3 F 10 G 12 Z Select candidate M2 Suspend M2 s constraints 9 13 00 copyright Brian …
View Full Document
Unlocking...