Condition Monitoring of Hydraulic Brake System Using Rough Set Theory and FUZZY Rough Nearest Neighbor Learning Algorithms

Vibration based continuous monitoring system for fault diagnosis of real time hydraulic brake system is presented in this study. Under different simulated fault conditions, the vibration signals were acquired by using a piezoelectric transducer. Data mining plays vital role in fault diagnosis study. The statistical parameters were extracted from the vibration signals for good and different fault conditions of the brakes system. Using the information reduction theory, the good features that discriminate different faulty conditions were formed. The selected features were then classified using the rough set method. The classification result of rough set theory for fault diagnosis of a hydraulic brake system was compared with the results of the fuzzy-rough nearest neighborhood (FRNN) algorithm. The comparative results have been discussed. Comparatively, roughest theory produced better classification accuracy with 96.9 %.