Smart factory-Bearing fault detection system (1)
|Monitored Machine: Industrial Pump Motor Assembly|
Mobilio has developed a wireless vibration and temperature data loggers with optional mobile machines, web platforms, and gateways to monitor the healthy status of industrial machines. The solution is installed in a large textile company; one of the monitoring points is the motor bearing at the couple ring end (CE).
During the monitoring, we monitored some changes in vibration and temperature patterns in terms of continuous data analysis, spectral and waveform analysis.
|Continuous analysisVibration and temperature trend analysis at the monitoring points generally provides the first indication of abnormal detection. Continuous monitoring (for example, if the sampling interval is 10 minutes), may show small slopes and rapid increases. Tools, such as moving averages, can help visualize trends in the collected data, soften slight variations, and emphasize large variations, as shown in the following images.|
|Note: This device has a magnetic base installed, which can make it difficult to detect these temperature changes. Nevertheless, the sensor has identified significant pattern changes. Screw mounts are recommended for more reliable data collection.|
Typically, the second step in an ideal evaluation is spectrum-based analysis. Proper resolution and dynamic range can be used to identify faults. The following image compares the two three-axis spectra. One was achieved after an anomaly in gray, with the other being achieved prior to a fault indicated by a blue, pink, and yellow curve.The increase in peaks associated with BPFI (Ball Pass Frequency Inside Race) and BPF (Ball Pass Frequency Outside Race) related to the rise of a carpet was mainly detected.
Another important tool that has information about the nature of the fault is the waveform in the time domain, especially if it has impulsive and abnormal characteristics. The main effect of the fault is that the RMS and peak-to-peak values are increased by waveform changes, such as the presence of periodic pulses. In addition, the signal on the new bearing generally has low amplitude, no pulse, and statistically near white noise. You can see the increase in these values in the following image: The first image represents the waveform before the fault, and the second image occurs after the fault is detected.