Three Phase Induction Motor's Stator Turns Fault Analysis Based on Artificial Intelligence

Three Phase Induction Motor's Stator Turns Fault Analysis Based on Artificial Intelligence

H. A. Taha Hussein (NAHDA University, Bani-Souf, Egypt), M. E. Ammar (Department of Electrical Power and Machines, Cairo University, Egypt) and M. A. Moustafa Hassan (Department of Electrical Power and Machines, Cairo University, Egypt)
Copyright: © 2017 |Pages: 19
DOI: 10.4018/IJSDA.2017070101
OnDemand PDF Download:
$37.50

Abstract

This article presents a method for fault detection and diagnosis of stator inter-turn short circuit in three phase induction machines. The technique is based on modelling the motor in the dq frame for both health and fault cases to facilitate recognition of motor current. Using an Adaptive Neuro-Fuzzy Inference System (ANFIS) to provide an efficient fault diagnosis tool. An artificial intelligence network determines the fault severity values using the stator current history. The performance of the developed fault analysis method is investigated using Matlab/Simulink® software. Stator turns faults are detected through current monitoring of a 2 Hp three phase induction motor under various loading conditions. Fault history is calculated under various loading conditions, and a wide range of fault severity.
Article Preview

1. Introduction

The use of induction machines is very common in all industries with a reported 90% utilization in electrical motor applications (Abdel et al., 2016; Joshi & Talange, 2016). The merits of the induction machine are its low price, ease of control and reliability. Investigating induction machines’ faults is crucial to minimize downtime and cost of damages as discussed in (Abdel et al., 2016; Joshi & Talange, 2016). Induction machine faults include winding faults, unbalanced stator and rotor, broken rotor bars, eccentricity and bearing faults. The failure due to stator and bearing faults breakdown accounts for 40~50% of total induction faults motivating prediction of stator faults to save high maintenance costs (Abdel et al., 2016; Joshi & Talange, 2016). The main goal of fault diagnosis is to determine the fault as early as possible and classify it correctly to facilitate localization. In order to ensure reliable diagnosis, advanced strategies have to be used. Fault diagnosis methods fall under three categories model based fault detection, process history based method and signal based method (Abdel et al., 2016; Gao et al., 2015). The model based technique requires a mathematical model that is derived from the physics of the system to describe the system in both quantitative and qualitative models (Gao et al., 2015). The process history based method depends on a sufficient history of measurements. Parameter estimation is based on the input output data where a mathematical equation could be realized using the measurements and decisions. Signal based models depend on online process system measurements. The use of high efficient sensors and Data Acquisition Card (DAC) to identify system fault is proposed in (Dalvand et al., 2016; Chen et al., 2014).

The main scope of fault diagnosis is to determine the fault as early as possible and classify it in a correct way to facilitate the localization of the fault. To verify the safe and reliable process, advance fault diagnosis strategies have to be used. Different methods of fault detection model base and knowledge base. Model based technique is considered to have a mathematical model that describes the process system. Mathematical model is derived from physics of the system. This model can describe the system in both quantitative and qualitative models.

Process history based methodology does not depend on any analytical relations referring to the physical system. The experimental or simulation measurements can give input output data for the process system. System characteristics are obtained based on the measurements as explained in (Mustafa et al., 2014). Expert system uses the knowledge based I/O data to provide the decision of fault diagnosing. It uses different algorithms to detect the faults (Neural Network, fuzzy system, Neuro Fuzzy …) as proposed in (Azar & Vaidyanathan, 2016; Zhu & Azar, 2015; Azar, 2010a, b; Jasim, 2010). Knowledge based data should contain variable states of process system for both faulty and healthy cases that will lead to accurate diagnosis. It uses the logical rules (IF condition THEN decision) rules to determine the fault as reviewed in (Joshi & Talange, 2016, Mustafa et al., 2014, Jasim, 2010).

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 6: 4 Issues (2017)
Volume 5: 4 Issues (2016)
Volume 4: 4 Issues (2015)
Volume 3: 4 Issues (2014)
Volume 2: 4 Issues (2013)
Volume 1: 4 Issues (2012)
View Complete Journal Contents Listing