Abstract
Recently, the functional representation approach has become a growing subarea of the model-based diagnostic approaches. It provides efficient mechanisms for diagnosing new faults. Sometimes, it suffers from computational complexity when repeating the same mechanism for old faults. Its performance can be improved by using a learning approach to avoid this repetition. It is found that the case-based reasoning approach performs well as a learning and reasoning technique. Simply, it diagnoses the old faults using the experience from its previous tests. This paper describes case-functional-based diagnostic system (CFDS) that introduces some new issues to simplify the complexity analysis of the functional-based diagnostic approach. Also, it proposes the integration of case- and functional-based reasoning approaches that improve their performance by getting the strength of both, and compensate the weakness of the other. CFDS can be applied for large varieties of domains. It has been applied for an uninterrupted power supply device, as a case study. It decreases the diagnostic process analysis, time, cost, and human errors. CFDS has satisfactory performance and user acceptance.