This paper is proposed a new method for identify stator fault in induction motor based on multinomial logistic regression analysis. A wavelet transform is used to calculate the value of high-frequency signals of motor electric current. The value of high frequency signal is then used as input variable of logistic regression to obtain the classification of the operating conditions that divided into a normal operation and symptom of damage. Three input variables (x1, X2, X3) which have been tested individually for modeling to identify the existence of fault. Those variables are obtained from three consecutive time period of current signal. Each period is 10ms. There is one input variable is X3 that no significant effect on the response variable, so that the simultaneous modeling of the variable is not included. Based on two input variables (x1 and X2) which are significantly affect response variables obtained, classification accuracy of stator fault identification is 77.5%.
|Title of host publication
|Proceeding - 2016 International Seminar on Intelligent Technology and Its Application, ISITIA 2016
|Subtitle of host publication
|Recent Trends in Intelligent Computational Technologies for Sustainable Energy
|Institute of Electrical and Electronics Engineers Inc.
|Number of pages
|Published - 20 Jan 2017
|2016 International Seminar on Intelligent Technology and Its Application, ISITIA 2016 - Lombok, Indonesia
Duration: 28 Jul 2016 → 30 Jul 2016
|Proceeding - 2016 International Seminar on Intelligent Technology and Its Application, ISITIA 2016: Recent Trends in Intelligent Computational Technologies for Sustainable Energy
|2016 International Seminar on Intelligent Technology and Its Application, ISITIA 2016
|28/07/16 → 30/07/16
- Multinomial Logistic Regression
- Stator Fault