Thesis
Fault management in networks incorporating Superconducting Cables (SCs) using Artificial Intelligence (AI) techniques.
- Creator
- Rights statement
- Awarding institution
- University of Strathclyde
- Date of award
- 2023
- Thesis identifier
- T16732
- Person Identifier (Local)
- 201988459
- Qualification Level
- Qualification Name
- Department, School or Faculty
- Abstract
- With the increasing penetration of renewable energy sources, the immense growth in energy demand and the ageing of existing system infrastructure, future power systems have started to face reliability and resiliency challenges. To mitigate these issues, the need for bulk power corridors which enable the effective sharing of the available power capacity, between countries and from remote renewable energy sources, is rendered imperative. In this context, the deployment of multi-layer Superconducting Cables (SCs) with High Temperature Superconducting (HTS) tapes have been considered as a promising solution towards the modernisation of power systems. As opposed to conventional copper cables, SCs are characterised by a plethora of technically-attractive features such as compact structure, higher current-carrying capability, lower losses, higher power transfer at lower operating voltages and over longer distances, and reduced environmental impact. The performance of SCs is mainly determined by the structure of the cable and the electro-magneto-thermal properties of the HTS tapes, accounting for the critical current, critical temperature and critical magnetic field. Particularly, during steady state conditions, HTS tapes operate in superconducting mode, providing tangible benefits to power system operation such as a current-flowing path with approximately zero resistance. However, under certain transient conditions (e.g., electric faults), when the fault current flowing through HTS tapes reaches values higher than the critical current, HTS tapes start to quench. The quenching phenomenon is accompanied by a rapid increase in the equivalent resistance and temperature of SCs, the generation of Joule heating and the subsequent reduction in fault current magnitudes. Consequently, the transition of SCs from superconducting state to resistive state, during transient conditions, introduces many variables in the fault management of such cable technologies. Therefore, in order to exploit the technological advantages offered by SC applications, accommodate their wide-scale deployment within future energy grids, and accelerate their commercialisation, the detailed evaluation of their transient response and the consequent development of reliable fault management solutions are vital prerequisites. On that front, one of the main objectives of this thesis is to provide a detailed fault signature characterisation of AC and DC SCs and develop effective and practically feasible solutions for the fault management of AC and High Voltage Direct Current (HVDC) grids which incorporate SCs. As the fault management (i.e., fault detection, fault location, and protection) of SCs has proven to be a multi-variable problem, considering the complex structure, the unique features of SCs, and the quenching phenomenon, there is a need for advanced methods with immunity to these factors. In this context, the utilisation of Artificial Intelligence (AI) methods can be considered a very promising solution due to their capability to expose hidden patterns and acquire useful insights from the available data. Specifically, data-driven methods exhibit multifarious characteristics which allow them to provide innovative solutions for complex problems. Given their capacity for advanced learning and extensive data analysis, these methods merit thorough investigation for the fault management of SCs. Their inherent potential to adapt and uncover patterns in large datasets presents a compelling rationale for their exploration in enhancing the reliability and performance of superconducting cable systems. Therefore, this thesis proposes the development of novel, data-driven protection schemes which incorporate fault detection and classification elements for AC and multi-terminal HVDC systems with SCs, by exploiting the advantages of the latest trends in AI applications. In particular this thesis utilises cutting-edge developments and innovations in the field of AI, such as deep learning algorithms (i.e., CNN), and state-of-the-art techniques such as the XGBoost model which is a powerful ensemble learning algorithm. The developed schemes have been validated using simulation-based analysis. The obtained results confirm the enhanced sensitivity, speed, and discrimination capability of the developed schemes under various fault conditions and against other transient events, highlighting their superiority over other proposed methods or existing techniques. Furthermore, the generalisation capability of AI-assisted schemes has been verified against many adverse factors such as high values of fault resistance and noisy measurement. To further evaluate the practical feasibility and assess the time performance of the proposed schemes, real-time Software In the Loop (SIL) testing has been utilised. Another very important task for the effective fault management of AC and DC SCs is the estimation of the accurate fault location. Identifying the precise location of faults is crucial for SCs, given their complex structure and the challenging repair process. As such, this thesis proposes the design of a data-driven fault location scheme for AC systems with SCs. The developed scheme utilises pattern recognition techniques, such as image analysis, for feature extraction. It also incorporates AI algorithms in order to formulate the fault location problem as an AI regression problem. It is demonstrated that the scheme can accurately estimate the fault location along the SCs length and ensure increased reliability against a wide range of fault scenarios and noisy measurements. Further comparative analysis with other data-driven schemes validates the superiority of the proposed approach. In the final chapter the thesis summarises the key observations and outlines potential steps for further research in the field of fault management of superconducting-based systems.
- Advisor / supervisor
- Tzelepis, Dimitrios
- Dysko Adam
- Resource Type
- DOI
Relazioni
Articoli
| Thumbnail | Titolo | Data caricata | Visibilità | Azioni |
|---|---|---|---|---|
|
|
PDF of thesis T16732 | 2023-10-16 | Pubblico | Scaricare |