Case-based fault diagnostic system

Abstract EN

Nowadays, case-based fault diagnostic (CBFD) systems have become important and widely applied problem solving technologies.

They are based on the assumption that “similar faults have similar diagnosis”.

On the other hand, CBFD systems still suffer from some limitations.

Common ones of them are: (1) failure of CBFD to have the needed diagnosis for the new faults that have no similar cases in the case library.

(2) Limited memorization when increasing the number of stored cases in the library.

The proposed research introduces incorporating the neural network into the case based system to enable the system to diagnose all the faults.

Neural networks have proved their success in the classification and diagnosis problems.

The suggested system uses the neural network to diagnose the new faults (cases) that cannot be diagnosed by the traditional CBR diagnostic system.

Besides, the proposed system can use the another neural network to control adding and deleting the cases in the library to manage the size of the cases in the case library.

However, the suggested system has improved the performance of the case based fault diagnostic system when applied for the motor rolling bearing as a case of study.