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

Theory · All theories
Evidence · no claims cited

Description¶

The article models fault diagnosis as two phases: partitioning components into a feasible and infeasible set given symptoms, then choosing a feasible member to test, repartitioning after each test result. Because humans have "difficulty in making simple.yes/no decisionsaboutthefeasibility ofeach component", the model uses fuzzy sets with "membership grades between 0.0 and1.0", logical operations for partitioning, and free parameters in membership functions matched to human performance. It was compared to Experiments One, Two, and Four.

Design Implications¶

Context¶

Requirements¶

• Free parameters within the membership functions must be fit to match the performance of the model and the human.

Constraints¶

• The model was developed and compared only against the results of Experiments One, Two, and Four, per the article.

Target Learners¶

• engineering or technical trainees performing fault diagnosis tasks

Target Learning Objectives¶

• troubleshooting and fault diagnosis skill

Claims¶

Related Theories¶

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

Examples¶

-

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