Search

Early detection and diagnosis of incipient faults is desirable for online condition assessment, product quality assurance, and improved operational efficiency of induction motors. In this paper, a speed-sensorless fault diagnosis system is developed for induction motors, using recurrent dynamic neural networks and multiresolution or Fourier-based signal processing for transient or quasi-steady-state operation, respectively. In addition to nameplate information required for the initial system setup, the proposed fault diagnosis system uses only motor terminal voltages and currents. The effectiveness of the proposed diagnosis system in detecting the most widely encountered motor electrical and mechanical faults is demonstrated through extensive staged faults. The developed system is scalable to different power ratings and it has been successfully demonstrated with data from 2.2-, 373-, and 597-kW induction motors. Index Terms--Fault diagnosis, induction motors, neural networks (NNs), signal processing.

Induction motors, both three and single phase, are used extensively for adjustable-speed drives' applications. These machines are structurally very robust and are a primary source of motive power and speed control where dc machines cannot be used. For closed-loop control of these machines, sensorless speed estimation is usually preferred. Among the current estimation techniques available for speed-sensorless induction motor drives, speed measurement based on rotor-slot-related harmonic detection in machine line current happens to be a prominent one. While these harmonics can be strong in certain kinds of machines, some other machines may exhibit very weak rotor slot harmonics that can be obscured by noise. Skewing, slot shapes and types, structural unbalances, etc., also have a prominent effect on the detectability of these harmonics. This paper attempts to investigate this problem based on the interaction of pole pairs, number of rotor bars, and stator winding. Although the analysis and experimental results have been mainly provided for three-phase squirrel-cage induction motors, single-phase and slip-ring induction motors have also been addressed. Further, it has been shown that eccentricity-related fault detection could also be easily accommodated with this kind of speed detection technique at no or negligible extra cost when certain motors are selected. Index Terms--Eccentricity, fault diagnosis, sensorless speed estimation.

A substantial portion of induction motor faults is eccentricity related. In practice, static and dynamic eccentricities happen to exist together. With this point in mind, an analytical approach to evaluate performance of a three-phase induction motor under mixed eccentric condition has been presented in this paper. Clear and step--by-step theoretical analysis, explaining completely the presence of certain harmonics in the line-current spectrum in presence of eccentricity, is discussed. More important, it is shown for the first time that a link exists between the low- and high-frequency elements of these harmonics. It is also shown that these high-frequency components are not very strong in all types of machines. These results will be useful in generating rules and laws to formulate online tools for machine-condition monitoring. Finite-element results to substantiate the inductance values used in the simulation are also included. The analysis is validated by the line-current spectrum of the eccentric machine obtained through simulation using modified winding-function approach (MWFA) and experimentation. Index Terms--Induction motor fault diagnosis, mixed eccentricity.

An online technique for monitoring the condition of the groundwall (GW) and phase-to-phase (PP) insulation for three-phase ac-machine stator windings is proposed in this paper. Insulation-condition indicators such as capacitance, dissipation factor (tan δ), and ac insulation resistance are calculated online based on the differential leakage-current measurements for each phase. A model for a three-phase ac-machine insulation system is derived and analyzed for both offline and online testing. Guidelines for interpreting the measured indicators for determining the: 1) overall condition and 2) significant degradation in the individual GW and PP insulation for each phase of the stator are given based on the analysis. Experimental results on a 15-hp induction motor under intentional GW and PP insulation degradation conditions show that incipient insulation degradation can be detected and classified with high sensitivity. [ABSTRACT FROM AUTHOR]

Early detection and diagnosis of incipient faults is desirable for online condition assessment, product quality assurance, and improved operational efficiency of induction motors. In this paper, a speed-sensorless fault diagnosis system is developed for induction motors, using recurrent dynamic neural networks and multiresolution or Fourier-based signal processing for transient or quasi-steady-state operation, respectively. In addition to nameplate information required for the initial system setup, the proposed fault diagnosis system uses only motor terminal voltages and currents. The effectiveness of the proposed diagnosis system in detecting the most widely encountered motor electrical and mechanical faults is demonstrated through extensive staged faults. The developed system is scalable to different power ratings and it has been successfully demonstrated with data from 2.2-, 373-, and 597-kW induction motors. [ABSTRACT FROM AUTHOR]