Data Mining of Transformer Frequency Response Analysis (FRA) Results for Characterisation and Interpretation

The infrastructure and equipment in the UK electrical power transmission system are encountering the issue of ageing, whilst the integration of renewable energy generations is expected to increase which continually challenge the capacity of accommodation renewable energy resource into the grid. A substantial proportion of transformers has been operated well beyond their designed lifetime, mainly due to low utilisation and good maintenance. Hence, it is essential to manage these aged assets in a safe and cost-effective manner to avoid catastrophic failures. One of the main challenges faced by asset managers is the unavailability of historical details dated to these transformers, including their designs. The analysis of frequency response measurements is effectively used for the determination of mechanical malfunction of transformers. This project focuses on data mining and mathematical modelling techniques to support asset managers to identify transformers with common design aspects and latent mechanical problems through the analysis of Frequency Response Analysis (FRA) measurement results, in addition to expert knowledge and forensic analysis. The research output is envisaged to allow incorporating more targeted management techniques to those individual transformer asset groups.

The tasks of this project are:

1. Collect all the available FRA measurements and establish a database in a
common format that enables automatic processing.

2. Develop an algorithm to process all the FRA measurements and identify similar
measurements and hence similar transformers.

The FRA measurement results to be used in the project will be from transformers
belonging to National Grid. Hence, the student will have the opportunity to spend a
considerable amount of time in National Grid as well as to work closely with their
transformer asset management. In order to achieve the tasks mentioned above, the student should conduct the project by using programming software, such as Matlab or C++ language, to develop new codes to implement the analysis of the FRA measurements. The conditions of aged transformer in service hopefully can be determined by analysing the results of FRA measurements according to the relevant standards of IEC and IEEE. In addition, the student will have the opportunity to attend local and international conferences during their study to share knowledge with other experts in the same area.