Rule-based model of fault diagnosis as selection among ordered rules-of-thumb¶

Theory · All theories
Evidence · no claims cited

Description¶

The article proposes that fault diagnosis involves "the use ofaset of rules-of-thumb (orheuristics)from which the human selects, using some type of priority structure." Based on Experiments Three, Five, and Six, an ordered set of twelve rules described Task One performance, matching subjects' tests 89% of the time, rising to 94% with a looser set of four rules; five rules gave a 38% match for Task Two. Rule rank ordering was affected by whether training was aided or unaided.

Design Implications¶

Context¶

Requirements¶

• A priority structure for selecting among the rules-of-thumb, as the article assumes.

Constraints¶

• Match rates are specific to the tasks reported: 89% (twelve rules) and 94% (four rules) for Task One, 38% (five rules) for Task Two.

Target Learners¶

• engineering or technical trainees performing fault diagnosis tasks

Target Learning Objectives¶

• troubleshooting and fault diagnosis skill

Claims¶

Related Theories¶

• Fuzzy set model of human fault diagnosis as partitioning and testing with graded membership

Examples¶

• Rate only poor tests in rule-based aiding: full quality ratings were counterproductive

Key Sources¶

• Rouse, W. B., Rouse, S. H., Hunt, R. M., Johnson, W. B., and Pelligrino, S. J. (1980). Human Decision-Making in Computer-Aided Fault Diagnosis. Technical Report 434, Coordinated Science Laboratory, University of Illinois. https://eric.ed.gov/?id=ED192743