Fault Detection and Diagnosis of Multi-Phase Induction Motor Drives Using MFRF Technique Balamurugan Annamalai Sivakumaran Thangavel Swaminathan Department of EEE Department of EEE Sathyabama Institute of Science and Technology Sasurie College of Engineering Chennai, India Thiruppur,India at.balamurugan@gmail.com sivakumaran1969@gmail.com . Abstract— In the dissertation, a hybrid technique based on detection and diagnosis of fault in multi-phase induction motor (IM) is performed. The present technique is the hybridization of Moth Flame optimization (MFO) and Random Forest algorithm (RFA) and it is named as MFRF method. The multiphase IM is evaluated under normal conditions in the initial period. The fault is maintained in multi-phase IM as well as characteristics of system are observed. In the defective period, signals are scaled, that may seen as waveforms are distorted. Distorted waveforms are made up of various frequency methods are required to represent as frequency of time domain as evaluation of failure. IM. The proposed technique is performed in MATLAB/Simulink platform. Implementation of established technique is contrasted to existing methods, like ANN, S-Transform and GBDT. The statistical measures are determined to demonstrate the successfulness of established technique, like precision, sensitivity and specificity, mean median and standard deviation. Keywords—Induction motor, fault detection, random forest algorithm, Moth flame optimization, ball bearing, inner race I. INTRODUCTION Being recommend the measures of economy and technology proven as three-phase machines control the speed in applications of market drive. But, current attempt has made with highpoint the benefits of various phase machines as well as get niche methods where they may finish to three-phase standards. Hence tolerance of failure implement as phase of redundant in single characteristics machines as various phase, industry has initially focused as maximum-reliability methods like aerospace, traction or systems of wind energy consists of machines to winding in three-phase as various phase [4]. To the applications of mind, various operations has focused in evolution of maximum-performance control methods of fault in machines as induction (IMs) as well as synchronous machines as permanent magnet (PMSMs) [5]. Diagnosis of failures in electrical machines may utilize to notice as emerging failures, and leads to rapid unplanned maintenance of corrective, short downtimes as well as reduction of side-effects in harmful [6]. Most types of failure (e.g. rotor bars broken or failure between turns, to name a little) must be considered in account and relate as detection of high times may allowed. Next, the objective of failure observation is implement as any tolerance methods to failure defined [7], detection of fault should focus on Open Phase fault (OPFs) [8] and times of detection as short as sufficient (generally lower than unique period of fundamental). Hence the detection of failure technique is contrasted to maximum execution as control of superior failure, namely; detection time as short (R1), locate the failure capacity (R2), methods of non- invasive and eliminate additional hardware (R3), eliminate complex methods to high cost of calculation (R4), be individualistic of parameter machine, control method and conditions of operating method (R5) [9-12]. Although it has few techniques of FD in literature on multiphase units, no one meet the techniques as aforementioned methods [13]. Method of detecting dissymmetry in stator resistance as seven-phase of IM is introduced [14]. Therefore, this technique depends on methods of control (R5 is insufficient). Requirement of R3 is breached [15-16], in case of extra measurements of voltage are required to notice short circuits between turns. Due to the view of failure detection technique is suggested [17] along to finite failure of tolerance control. To implement the viewer as cost of calculation based on parameter of machines. Subsequently, requirements of R4 and R5 are not met to this technique. To view the point as superior- failure control, OPF has situation of favorable studied failure nevertheless methods of control [17-20]. Hence, there is no technique to detect this way of failure that meets the conditions mentioned above. II. RECENT RESEARCH WORKS: A BRIEF REVIEW Several investigation methods in bibliography which depends on diagnosis of failures in induction machines of various phase utilized as various techniques and various points of view. Here a part of the works is examined. Due to evaluation of signal as quaternion, J. Contreras-Hernandez et al. [21] has presented a new engine fault detection method. To maintain the rotation of quaternion and implies statistics rotation of quaternion, like mean, the shadows of clusters and the prominence of clusters to obtain their functions, and employ to allocate the state of motor utilize the algorithm as classification of tree. Its strategy was experimentally validated and contrasted to another technique demonstrate as organization technique as detection of characteristics and recognition and allocation of engine failures. M. Singh and A. Shaik [22] have used the detection, allocation and position of defective bearings as induction motor in three-phase employ as Vector Stock-well transformer and support machine. Stock- well transformation was performed to signals of stator current remove the series as characteristics time and domain frequency. Due to the score of Fisher rating the place uncorrelated maximum-ranking characteristics was selected. These characteristics were change to allocate failures, like ball, cage and outer stroke failures, employ as support vector machine. The characteristic of Stock-well transformation was utilized after the identification of failures, to place the bearing as defective, i.e, fan side or motor burden side. A. Background of the Research Work Analysis of current investigation displays the observation of failures in motor of induction as significant factor of contribution. Earlier diagnosis of failure is most significant to eliminate the effects of catastrophic in 978-1-7281-6368-0/20/$31.00 ©2020 IEEE 2020 5th International Conference on Devices, Circuits and Systems (ICDCS) 247 2020 5th International Conference on Devices, Circuits and Systems (ICDCS) 978-1-7281-6368-0/20/$31.00 ©2020 IEEE 10.1109/ICDCS48716.2020.243590