Comparison Of Machine Learning Methods In Predıctive Maintenance Applications Of Electric Motors
Berk Kahyaoğlu1*, Nuri Berkcan Sarıel2, Erdinç Berkin Konca3
1Volt Electric, İzmir, Turkey
2Volt Electric, İzmir, Turkey
3Volt Electric, İzmir, Turkey
* Corresponding author: berk.kahyaoglu99@hotmail.com
Presented at the 6th International Symposium on Innovative Approaches in Smart Technologies (ISAS-WINTER-2022), Online, Turkey, Dec 08, 2022
SETSCI Conference Proceedings, 2022, 14, Page (s): 45-48 , https://doi.org/10.36287/setsci.5.2.010
Published Date: 22 December 2022 | 1835 19
Abstract
Three-phase asynchronous motors are widely used in many areas of industry. Unplanned malfunctions in the motor can cause the entire system to become inefficient. Detecting and intervening the malfunctions before they grow has a direct impact on the life of the motor and contributes positively to the operating economy. Therefore, it is important to detect motor failures in advance and perform predictive maintenance. In this study, the vibration data of a three-phase asynchronous motor of Volt Electric Motors under full load was measured with a motion sensor. The created data set was analyzed by processing with Polynomial Regression, Spectral Analysis, ARIMA and Artificial Neural Network models. An attempt was made to predict future data. The artificial Neural Network model gave the most effective result among these models. First, classes were created with the health data of the motor and the corrupted versions of this data at certain rates. Then, the model was established and the training was carried out. The model classified the test data with 96% accuracy. By looking at this classification, it has been concluded that it can be determined whether the motor is defective or not. Four different methods were analyzed and the results were shared.
Keywords - Electric Motor, Predictive Maintenance, Artificial Intelligence, Machine Learning, Artificial Neural Network
References
[1] Gamze Kaynak, “Kestirimci Bakım Planlaması için Zaman Serisi Modellerine Dayalı Arıza Zamanı Tahmini”, 2022.
[2] Hasan Bora Denli , “Kestirimci bakım ve uygulamalarının geliştirilmesi”,2007
[3] Wei Wu, Jingtao Hu and Jilong Zhang, “Prognostics of Machine Health Condition using an Improved ARIMA- based Prediction method”, 2007.
[4] Nadhem Nedhal Abdo Qaid, MRI Brain Tumor Classification Using Texture Features with Artifical Neural Network and Deep Learning