Power asset management by using statistical analysis and long short-term memory networks for old-immersed transformer | |
| Author | Noparada Sutthichackr |
| Call Number | AIT Thesis no.ET-21-12 |
| Subject(s) | Electric power distribution--Management Statistics--Computer programs--Analysis |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Energy |
| Publisher | Asian Institute of Technology |
| Abstract | Power asset management is not a new concept. It's been with us since the beginning of the project. The asset management system should be consistent with the goal and objective of the company to develop the organization to be successful. Maintenance of important assets is a significant part of asset management in the distribution network. General problems of asset maintenance are caused by a lack of maintenance planning or decision-making even if the equipment is repaired according to usage history documents. In addition, statistical analysis is one of several asset maintenance techniques, which can be used for monitoring machine conditions by calendar time and for failure forecasting. Furthermore, there are many devices in the electrical power system, but the most important equipment is the power transformer because it is the main power supply and its working status is related to stability and safety. Moreover, the DGA method is a solution to detects incipient transformer faults to prevent further damage. However, DGA cannot predict the incipient fault in real-time, and its cost testing is expensive. Thus, machine learning was selected to predict the incipient fault in the transformer based on DGA data. It makes transformer will be achieved better mechanism for health assessment. This study aims to use statistical analysis to identify the most outage substation and use LSTM for the transformer to predict the incipient fault in real-time in the institute. This thesis focuses only on the distribution power system in AIT and DGA data from literature and TU transformer under controlled by Property and Sports Management Thammasat University. The result is presented that there are 15 substations in AIT. Nevertheless, substation 9 has the highest expected failure, which is approximately 23 failures, which means after the last failure occurs, there are 23 failures in the next year. In addition, the highest accuracy model from LSTM is 40 cells and 7 layers, and it is used to test with DGA data from literature and TU transformer which the accuracy score is 95 and 97 percentages, respectively. The importance of this study is the power system’s reliability in this area increased. |
| Year | 2021 |
| Type | Thesis |
| School | School of Environment, Resources, and Development (SERD) |
| Department | Department of Energy and Climate Change (Former title: Department of Energy, Environment, and Climate Change (DEECC)) |
| Academic Program/FoS | Energy Technology (ET) |
| Chairperson(s) | Singh, Jai Govind |
| Examination Committee(s) | Weerakorn Ongsakul;Roy, Joyashree |
| Scholarship Donor(s) | Royal Thai Government Fellowship |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2021 |