Powertrain Fault Diagnosis of Manufacturing Processes by Means of Servo Drives

During cyclical manufacturing processes, powertrain and process faults such as friction, distributed damage and material fatigue can occur. The early detection of faults through condition monitoring allows to minimize production losses and cost. Here, a condition monitoring approach is presented, which detects irregular deviations from the wished operation via the servo drive and does classify them. The approach distinguishes into signal analysis, characterization and classification. The analysis is summarized by characterization in representative values. The subsequent classification compares the characterization values of the measurement cycle with a reference cycle and classifies these into different classes, e.g. by pattern recognition. Various fault characteristics were simulated on a HiL test bench and the presented procedure was applied to the measurement variables rotational speed and torque-generating current of the servo drive. It can be stated that a predominant share of test cases is appropriately classified.