Application of Advanced Vibration Monitoring Systems and Long Short-Term Memory Networks for Brushless DC Motor Stator Fault Monitoring and Classification

Abstract

In this research, electric motors faults and their identification is reviewed. Brushless direct-current (BLDC) motors stator fault identification using long short-term memory neural networks were analyzed. A proposed method of vibration data acquisition using cloud technologies with high accuracy, feature extraction using spectral entropy, and instantaneous frequency and standardization using mean and standard deviation was reviewed. Additionally, model training with raw and standardized data was compared. A total model accuracy of 97.10 percent was achieved. The proposed methods could successfully identify the motor stator status from normal, to loss of stator winding imminent and arcing, and lastly to open circuit in stator winding—motor needing to stop immediately—by using gathered data from real experiments, training the model and testing it theoretically.

A Technique for Frequency Converter-Fed Asynchronous Motor Vibration Monitoring and Fault Classification, Applying Continuous Wavelet Transform and Convolutional Neural Networks

“…The main motor parameters were: Siemens asynchronous motor, 1.1 kW power rating with 2.6 A current, connected as star, cosφ-0.81 and 1415/min revolutions at 50 Hz frequency; efficiency class-IE1-75%. The data acquisition part was the same as in [32] and is shown in Figure 2.…”

“…The communication was done using Profinet and recorded data was uploaded to the cloud by the Siemens measurement PLC using Siemens MindConnect Nano gateway, which was accessible by laptop to retrieve and process. More detailed specifications of the equipment can be found in Tables 1 and 2 of article [32]. Siemens monitoring PLC, marked 2, and the "Mindsphere" gateway (MindConnect Nano), marked 1, in Figure 2.…”

“…The main motor parameters were: Siemens asynchronous motor, 1.1 kW power rating with 2.6 A current, connected as star, cosϕ-0.81 and 1415/min revolutions at 50 Hz frequency; efficiency class-IE1-75%. The data acquisition part was the same as in [32] and is shown in Figure 2. In Figure 1, a Siemens asynchronous motor can be seen (marked as number 1), with vibration sensor installed into the bearing housing plate and marked as number 2.…”

“…Test results are described in Section 3.2. Confusion matrix acquisition after training is described more in detail in [32].…”

“…In this research, a similar approach will be used but using more complex models to simplify and accelerate the training process, also increasing accuracy. This is a continuation of research [32], in pursuit of creating an all-around self-identifying and highly applicable electric motor monitoring model. While in [32], a brushless DC motor was used, it was decided to move to a more widely used motor in production-an asynchronous motor-and it was decided to change the model in order to simplify it.…”