Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems
Condition monitoring of electrical machines is a field which continuously demands improvement, especially as the tasks performed by these machines become more demanding and complex. Presently, the well established condition monitoring techniques are based on steadystate analysis of various diagnosti...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Department of Electrical Engineering
2024
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613780671725568 |
|---|---|
| access_status_str | Open Access |
| author | Barendse_PS |
| author2 | Professor P. Pillay |
| author_browse | Barendse_PS Professor P. Pillay |
| author_facet | Professor P. Pillay Barendse_PS |
| author_sort | Barendse_PS |
| collection | Thesis |
| description | Condition monitoring of electrical machines is a field which continuously demands improvement, especially as the tasks performed by these machines become more demanding and complex. Presently, the well established condition monitoring techniques are based on steadystate analysis of various diagnostic variables (i.e. current, voltage, power, etc.). These steadystate techniques are currently being applied to most machine-related applications, however it is well known that machine diagnostic variables behave typically as non-stationary signals. Therefore, it is difficult distinguishing fault conditions from normal operating conditions using steady-state analysis techniques. The most popular of these steady-state techniques employs the Fourier Transform as the analysis tool. Although there are numerous diagnostic variables which may be utilized for condition monitoring purposes, e.g. thermal and vibration monitoring have been quite popular, the most recent research is directed towards electrical monitoring of the motor. The thesis examines the use of two time-frequency domain signal processing tools in its application to condition monitoring of electrical machine drive systems. The mathematical and signal processing tools which are explored are wavelet analysis and a non-stationary adaptive signal processing algorithm. Four specific applications are identified for the research. These applications were specifically chosen to encapsulate important issues in condition monitoring of variable speed drive systems. The main aim of the project is to highlight the need for fault detection during machine transients and to illustrate the effectiveness of incorporating and adapting these new class of algorithms to detect faults in electrical machine drive systems during non-stationary conditions. The first application investigates the use of an adaptive algorithm for fault detection of inverterfed permanent magnet (PM) machines during non-stationary conditions using the stator currents. The detection technique is based on the non-stationary adaptive signal processing algorithm which has been cascaded to extract fault associated sinusoids using the current signals. The machine faults examined include permanent magnet damage, inter-coil shorts and static eccentricity. The results indicate that the algorithm's predictive ability is capable of extracting fault information under non-stationary operating conditions. This second application examines the collaborative use of the non-stationary adaptive algorithm and the Wavelet Transform for the detection of mechanical imbalances in inverter-fed induction machines. The fault frequency components for induction machines are dependant on two variables, unlike the single variable dependence by permanent magnet machines. This poses too much of a challenge for the predictive capability of the cascaded adaptive algorithms proposed for the PM case and an alternate approach is explored. The Wavelet Transform is incorporated into the final detection scheme and it is shown that imbalanced faults in inverter-fed induction machines can be identified by decomposing transient inrush currents. This is significant, since most fault detection schemes fail or operate at a diminished level when drives are connected to the machine. The detection algorithm is load dependent, however it only requires a minimum load of approximately 30% at the desired speed. This is a considerable improvement from steady-state analysis techniques where heavier load conditions are required, particularly in the case of inverter-fed machines. The third application investigates the detection of inter-tum stator faults in doubly-fed induction generators (DFIG), which are typically used in wind generator applications. The decision to analyze the system is based on their popularity in utility wind generator applications and because these generators can operate in both synchronous and asynchronous modes. A new nonstationary method of detecting inter-tum stator faults is proposed. The proposed fault detection method is a combination of Extended Park's Vector Approach and the non-stationary adaptive algorithm. The new method shows that inter-tum stator faults can unambiguously be identified during non-stationary conditions while also providing insight into the severity of the fault. In the previous applications, the faults stem from machine failure, however it is important for this research to incorporate at least one fault, which is not machine related, but which could have an impact on the health of the machine and drive. The doubly-fed induction generator proves to be an ideal candidate, whereby a sag in the grid voltage could cause large induced currents within the rotor circuit. This would be catastrophic to the power electronics and machine. The voltage sag therefore needs to be detected almost immediately. The sag detection technique is based on the non-stationary adaptive algorithm and is shown to detect sags faster than previous sag detection techniques. A simple mitigation strategy is employed to illustrate the effectiveness of diverting the excess energy once the sag is detected. The proposed strategy is simple and easy to incorporate as part of an existing control system. |
| format | Thesis |
| id | oai:open.uct.ac.za:11427/39901 |
| institution | University of Cape Town (South Africa) |
| language | eng |
| last_indexed | 2026-06-10T12:41:35.593Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UCTD — University of Cape Town Open Access Repository |
| publishDate | 2024 |
| publishDateRange | 2024 |
| publishDateSort | 2024 |
| publisher | Department of Electrical Engineering |
| publisherStr | Department of Electrical Engineering |
| record_format | dspace |
| source_str | UCTD — University of Cape Town Open Access Repository |
| spelling | oai:open.uct.ac.za:11427/39901 The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems Barendse_PS Professor P. Pillay Electrical Engineering Condition monitoring of electrical machines is a field which continuously demands improvement, especially as the tasks performed by these machines become more demanding and complex. Presently, the well established condition monitoring techniques are based on steadystate analysis of various diagnostic variables (i.e. current, voltage, power, etc.). These steadystate techniques are currently being applied to most machine-related applications, however it is well known that machine diagnostic variables behave typically as non-stationary signals. Therefore, it is difficult distinguishing fault conditions from normal operating conditions using steady-state analysis techniques. The most popular of these steady-state techniques employs the Fourier Transform as the analysis tool. Although there are numerous diagnostic variables which may be utilized for condition monitoring purposes, e.g. thermal and vibration monitoring have been quite popular, the most recent research is directed towards electrical monitoring of the motor. The thesis examines the use of two time-frequency domain signal processing tools in its application to condition monitoring of electrical machine drive systems. The mathematical and signal processing tools which are explored are wavelet analysis and a non-stationary adaptive signal processing algorithm. Four specific applications are identified for the research. These applications were specifically chosen to encapsulate important issues in condition monitoring of variable speed drive systems. The main aim of the project is to highlight the need for fault detection during machine transients and to illustrate the effectiveness of incorporating and adapting these new class of algorithms to detect faults in electrical machine drive systems during non-stationary conditions. The first application investigates the use of an adaptive algorithm for fault detection of inverterfed permanent magnet (PM) machines during non-stationary conditions using the stator currents. The detection technique is based on the non-stationary adaptive signal processing algorithm which has been cascaded to extract fault associated sinusoids using the current signals. The machine faults examined include permanent magnet damage, inter-coil shorts and static eccentricity. The results indicate that the algorithm's predictive ability is capable of extracting fault information under non-stationary operating conditions. This second application examines the collaborative use of the non-stationary adaptive algorithm and the Wavelet Transform for the detection of mechanical imbalances in inverter-fed induction machines. The fault frequency components for induction machines are dependant on two variables, unlike the single variable dependence by permanent magnet machines. This poses too much of a challenge for the predictive capability of the cascaded adaptive algorithms proposed for the PM case and an alternate approach is explored. The Wavelet Transform is incorporated into the final detection scheme and it is shown that imbalanced faults in inverter-fed induction machines can be identified by decomposing transient inrush currents. This is significant, since most fault detection schemes fail or operate at a diminished level when drives are connected to the machine. The detection algorithm is load dependent, however it only requires a minimum load of approximately 30% at the desired speed. This is a considerable improvement from steady-state analysis techniques where heavier load conditions are required, particularly in the case of inverter-fed machines. The third application investigates the detection of inter-tum stator faults in doubly-fed induction generators (DFIG), which are typically used in wind generator applications. The decision to analyze the system is based on their popularity in utility wind generator applications and because these generators can operate in both synchronous and asynchronous modes. A new nonstationary method of detecting inter-tum stator faults is proposed. The proposed fault detection method is a combination of Extended Park's Vector Approach and the non-stationary adaptive algorithm. The new method shows that inter-tum stator faults can unambiguously be identified during non-stationary conditions while also providing insight into the severity of the fault. In the previous applications, the faults stem from machine failure, however it is important for this research to incorporate at least one fault, which is not machine related, but which could have an impact on the health of the machine and drive. The doubly-fed induction generator proves to be an ideal candidate, whereby a sag in the grid voltage could cause large induced currents within the rotor circuit. This would be catastrophic to the power electronics and machine. The voltage sag therefore needs to be detected almost immediately. The sag detection technique is based on the non-stationary adaptive algorithm and is shown to detect sags faster than previous sag detection techniques. A simple mitigation strategy is employed to illustrate the effectiveness of diverting the excess energy once the sag is detected. The proposed strategy is simple and easy to incorporate as part of an existing control system. 2024-06-19T07:06:45Z 2024-06-19T07:06:45Z 2007 2024-06-18T12:14:14Z Thesis / Dissertation Doctoral PHD http://hdl.handle.net/11427/39901 eng application/pdf Department of Electrical Engineering Faculty of Engineering and the Built Environment |
| spellingShingle | Electrical Engineering Barendse_PS The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| thesis_degree_str | Doctoral |
| title | The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| title_full | The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| title_fullStr | The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| title_full_unstemmed | The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| title_short | The application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| title_sort | application of advanced signal processing techniques to the condition monitoring of electrical machine drive systems |
| topic | Electrical Engineering |
| url | http://hdl.handle.net/11427/39901 |
| work_keys_str_mv | AT barendseps theapplicationofadvancedsignalprocessingtechniquestotheconditionmonitoringofelectricalmachinedrivesystems AT barendseps applicationofadvancedsignalprocessingtechniquestotheconditionmonitoringofelectricalmachinedrivesystems |