Dragan Matic

Support vector machine classifier for diagnosis in electrical machines

Highlights? System for broken bar detection for wide slip range. ? Detection based on measuring only a motor current. ? No need for mathematical models, classifier is trained on acquired data. ? Detection of broken bar at low slip, where classical broken bar detection classifiers are not applicable. ? Reliable, mobile, and cost effective system that can be successfully applied in real working environment. This paper presents a support vector machine classifier for broken bar detection in electrical induction machine. It is a reliable online method, which has high robustness to load variations and changing operating conditions. The phase current is only physical value to be measured. The steady state current is analyzed for broken bar fault via motor current signature analysis technique based on Hilbert transform. A two dimensional feature space is proposed. The features are: magnitude and frequency of characteristic peak extracted from spectrum of Hilbert transform series of the phase current. For classification task support vector machine is used due to its good robustness and generalization performances. A comparative analysis of linear, Gaussian and quadratic kernel function versus error rate and number of support vectors is done. The proposed classifier successfully detects a broken bar in various operational situations. The proposed method is sufficiently accurate, fast, and robust to load changes, which makes it suitable for use in real-time online applications in industrial drives.

Artificial neural networks broken rotor bars induction motor fault detection

Paper deals with application of online rotor broken bar fault detection via artificial neural networks. Fault can be detected by monitoring abnormalities of the spectrum amplitudes at certain frequencies in the motor current spectrum. These discriminative features are used for training of feed-forward backpropagation artificial neural network. Trained network is capable to successfully classify induction motor rotor condition. Results are presented in tables and figures.

Induction motor broken rotor bar detection using vibration analysis — A case study

Early fault diagnosis and condition monitoring can reduce the consequential damage and breakdown maintenance, prolong the machine life and increase the performance of industrial systems. This paper describes a real fault detection problem of a high-power (3.2 MW) induction motor driving pumps in a heating plant. Steady-state vibration signals were collected and some characteristic low- and high- frequency features were observed, resulting in broken rotor bar diagnosis. The obtained results were verified by disassembling the motor, which proved that this particular technique can be successfully applied in induction motor fault detection.

Artificial Neural Networks Eccentricity Fault Detection of Induction Motor

This paper deals with eccentricity fault detection in a induction motor via artificial neural networks. Discriminative features are extracted from magnitudefrequency plot of line current spectra at characteristic frequencies. Based on this data, training and test sets for used artificial neural networks are made. Feedforward and radial basis function neural networks are used for tasks of rotor condition classification. Well trained artificial neural networks are capable to successfully classify rotor condition at medium and full shaft load for choosen features. Simple structure and implementation made them suitably for practical usage.

Broken bar detection using current analysis — A case study

This paper covers a case study of broken bar detection for 3.15 MW motor in a thermal power plant application. The motor current is measured in one phase. Feature extraction is based on transient and steady state analysis. Hilbert and Wavelet transforms are used to extract broken bar features. To discuss rotor condition in time domain skewness and kurtosis of current envelope are also considered. Low shaft-load conditions are present. In case of high-voltage, high-power induction motor reliable broken bar detection is possible when contemporary digital signal processing techniques are used.

Design of support vector machine classifier for broken bar detection

This paper proposes method for broken bar detection in induction motors at very low slip. The proposed method consists of extracting reliable discriminative feature from a steady state one-phase current signal and design of optimal classifier via a support vector machine. The fault related features are extracted from frequency spectra of a modulus of a motor phase current Hilbert transform series. The features are fed to the support vector machine input and the output indicates rotor condition in respect of broken bar appearance independently of a slip value. Tests are conducted on 1.1kW two poles induction machine in an industrial environment. It is shown that proposed method is accurate, fast, reliable, not hardware costly.

GA optimization of PI controller in MRAS structure for induction motor speed estimation

This paper represents genetic algorithm optimization of PI controller parameters in model reference adaptive system structure for speed estimation of induction motor. Experiments were taking out via dSPACE DC1104 digital control card at laboratory experimental drive. For needs of simulation and model building Matlab/Simulink, software was used. Experimental results show that known theoretical and experimental foundations published by various authors support used methodology.

The estimation of iron losses in a non-oriented electrical steel sheet based on the artificial neural network and the genetic algorithm approaches

Cold rolled non-oriented (CRNO) electrical steel sheets are soft ferromagnetic materials which are commonly used for electromagnetic core design for AC rotating electrical machines. When these materials are exposed to time-varying magnetic fields, the iron losses occur. These losses represent the power dissipated in the ferromagnetic material and they are dependent upon the frequency and magnetic flux density level of the applied time-varying magnetic field. In order to achieve high-efficiency electrical machines, especially at high operating frequencies and magnetic flux density levels, iron losses should be kept as low as possible. This imposes the need for more accurate iron losses models, but also for fast and reliable estimation techniques. This paper considers the applications of an artificial neural network (ANN) and a genetic algorithm (GA), based on the classical iron losses separation formulation for a fast estimation of the specific iron losses in CRNO electrical steel sheet grade M530-50A over a wide frequency and magnetic flux density range. Iron losses measurement data, provided by the manufacturer, are used to calibrate the iron losses models. The approaches were verified using the manufacturers measurement data. Acceptable accuracy was obtained.