INTELLIGENT CONDITION ASSESSMENT AND REMAINING USEFUL LIFE–ORIENTED MAINTENANCE PREDICTION FOR TRACTION MOTOR MECHANICAL FAULTS

Abstract

This study proposes an interpretable fuzzy-logic framework for the integrated condition assessment of railway traction motors using bearing temperature, vibration velocity, acoustic noise, and magnetic-field variation. Three linguistic states represent each diagnostic variable, and their combinations are processed using a Mamdani fuzzy inference system comprising 81 rules. The proposed framework converts heterogeneous sensor information into three operational decision classes: continued operation, maintenance required, and protective intervention. The model was implemented in MATLAB/Simulink and linked to a maintenance-planning module that calculated the distance remaining to the scheduled maintenance limit. Scenario-based verification using nine representative combinations of normal, intermediate, and critical parameter values showed that the fuzzy-system decisions were consistent with the predefined reference classes. The results confirm the logical consistency of the rule base and demonstrate the model’s ability to support interpretable condition-oriented decisions. The proposed approach provides a basis for online traction motor monitoring and maintenance planning; however, further validation using long-term locomotive measurements and documented fault records is required before operational deployment.