Identification of anomalies in highly-integrated electric drives by secondary excitation mechanisms

Series production of electric drives tends towards highly-integrated topologies. All components ranging from the inverter over the electric engine to the reduction gear are mounted in a single housing. The identification of faults by measuring structure-borne noise exhibits new challenges as each vibration of a part is dependent on its surroundings. A metric for calculating a measure for anomaly regarding the mean vibration signature is compared to known metrics. The reduced difference-correlation metric is exploited further to address anomalies across different order tracks, consequently scanning spectrograms for anomalous coincidences. The method is evaluated by two classes of faults (electrical and mechanical). The faults can not be distinguished by elementary order analyses as they match in their primary excitation. Including the view of secondary fault excitations each fault can be classified. The interconnection between different emerging orders is clearly observable and the faults can be classified.