Condition Diagnosis on a Gear Motor with Neural Networks Utilizing Vibration Data and the Current Values for Efficiency Enhancement

In many production-related applications, electromechanical drive units with gear motors are not maintained in a condition-based manner. The worsening system conditions lead to deterioration in efficiency and effectiveness with an increasing operating time. Vibration measurements, as well as the current signals of an electric motor, offer the possibility to monitor the condition of gear motors. In laboratory environments, the generation of fault conditions is difficult because this often requires long operating times. For this reason, the measurements are carried out on a gear motor in new condition and on a field return, which was used under real operating conditions in production. Different operating conditions of the drive units represent the conditions in real production environments and have an influence on the state variables “vibration” and “current”. Therefore, the different operating conditions must be considered to ensure reliable monitoring of the system state. To enable a fault detection with a multi-input artificial neural network (ANN), preprocessing of the recorded vibration data with an envelope demodulation is necessary for highlighting the relevant signal components. Only the maximum current values per operating and system condition are considered. The consideration of the current and vibration signals improves the accuracy of condition diagnosis.