Khan, Md. Abdesh Shafiel Kafiey (2006) A wavelet packet transform based on-line technique for the protection of three-phase interior permanent magnet motors. Masters thesis, Memorial University of Newfoundland.
- Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.
Interior permanent magnet (IPM) synchronous motors are attractive for a variety of applications because of their high electromagnetic torque per permanent-magnet volume, high efficiency, low noise and low cost with a high power factor. However, interior permanent magnet motors are subject to time-to-time transients primarily due to faults and switches of loads. As a result, it is very important to understand the IPM motor's response to any potential fault condition in order to prevent fault-induced damage to the motors or to the connected loads. There have been many proposed techniques to carry out the protection of interior permanent magnet motors. These techniques include the harmonic content analysis of line currents using the fast Fourier transform (FFT), dq modeling of !PM motors relative to the faults, and pattern recognition of fault currents using the artificial intelligence techniques such as expert systems (ES), artificial neural networks (ANN), fuzzy logic systems (FLS), genetic algorithm (GA) and support vector machine (SVM). However, output signals of the running machine often contain nonstationary components and the FFT-based methods are not suitable for the non-stationary signal analysis. Moreover, there have been few efforts in the experimental implementation of dq modeling based protection of interior permanent magnet motors. Furthermore, pattern recognition techniques require a large number of data files for offline training and have computational complexity, which may affect the reliability and speed of operation of these protection techniques. In this work, a wavelet packet transform based algorithm for the protection of the interior permanent magnet motor is developed, implemented and tested on-line on two types of interior permanent magnet motors using the DS 11 02 digital signal processor board. In the proposed algorithm, a signature analysis method is used for extracting the features of fault currents. For this purpose, three-phase line currents of different faulted and normal unfaulted conditions are acquired through the three-channel A/D converters of the DS1102 digital signal processor board, and then these are decomposed off-line up to the second level of resolution of the wavelet packet transform. The second level high frequency details (dd²) only show the distinctive features between faulted and normal unfaulted (healthy) currents. So, the proposed wavelet packet transform based algorithm is dependent upon the identification of high frequency components or coefficients in the second level high pass filter output. The output of cascaded filter banks associated with the selected mother wavelet and number of levels of resolution produce details and approximations of the original signal, which are related to high frequency and low frequency components, respectively. An experimental setup is developed for collecting line currents of different faulted and normal (unfaulted) conditions in order to select the optimum mother wavelet and optimal number of levels of resolution in addition to off-line testings of the proposed algorithm. It is found that the mother wavelet 'db3' of the Daubechies family and the second level of resolution are adequate enough to discriminate fault currents from the normal unfaulted currents. The off-line testing of the proposed algorithm is performed in the MATLAB environment on the collected data from both types of interior permanent magnet motors. An experimental setup for detecting disturbances incorporating also the protection circuit is developed to accommodate on-line testings on both types of IPM motors. In all the tests carried out, the proposed algorithm identified disturbances properly and initiated a trip signal almost at the instant or within one cycle of the fault occurrence. The on-line testing results are consistent with those obtained from off-line testings. The real-time testing of the proposed algorithm is also extended on the inverter-fed interior permanent magnet motor to demonstrate the universality of the proposed wavelet packet transform (WPT)-based protection algorithm for either the line-fed or inverter-fed three-phase interior permanent magnet motors.
|Item Type:||Thesis (Masters)|
|Additional Information:||Bibliography: leaves 163-172.|
|Department(s):||Engineering and Applied Science, Faculty of|
|Library of Congress Subject Heading:||Electric motors, Synchronous; Neural networks (Computer science); Permanent magnet motors.|
Actions (login required)