Intelligent real-time fault diagnosis: Methodology and application

Abstract

This paper presents a methodology for application of Intelligent Systems to Fault Diagnosis for Realtime applications. Following the introduction, design of the Intelligent Diagnostic System is discussed. A case study is then, described and the design methodology is applied to the case under investigation. Results of the application of case study are discussed in detail. In this approach an Intelligent System of Artificial Neural Networks ISANN has been trained to learn malfunctions of the test system as well as healthy or standard operational status. ISANN is capable of testing the system under investigation periodically, on a prescheduled plan or on request. Upon completion of the test, condition of operation will be identified and reported with accuracy as being either under normal operation status, or faulty. In the event of a system fault, the diagnostic test identifies the exact cause of failure and the degree of system deterioration with a precision in excess of 98%. The study provides provisions for optimization of training set resulting in an expedited training stage of ISANN.