A Combinational-Logic Method for Electric Vehicle Drivetrain Fault Diagnosis

This paper presents a combinational-logic-based approach for identifying faults that could occur in the drivetrain of an electric vehicle (EV). A real-time simulation model of an EV is used to study the behavior of available sensor signals and control commands, with specific measured quantities, during vehicle operation. Those quantities could be, but are not limited to, mean values of the phase currents, DC bus current, traction motor speed (if available), in addition to control command quantities. Focus is given to these quantities which are abundant in vehicle inverters and controllers for no added sensing costs. Those quantities carry information of offsets and disturbances that could occur under faulty operating conditions compared to nominal operation, and are thus observed and studied under these conditions. Using such information, a method is developed using simple combinational logic and thresholds to diagnose a fault occurring at any time during a vehicle drive cycle. By combining features of measured quantities that behave similarly irrespective of when the fault occurs during a driving cycle, the proposed method is fault-time-insensitive. The proposed method is presented and validated by real-time simulations to capture over 20 different faults injected at different drive cycle times and in different drivetrain components-electric machine, inverter, transmission, and sensors. Results show that the proposed method is able to robustly and successfully diagnose different faults irrespective of when would they occur.