Determining boundary conditions of transformer station clusters by Inductive Fuzzy Classification

The energy sector is changing from being unimaginative and lethargic to focused, exciting and vigorously dynamic. External influences in this sector are constantly increasing: global warm- ing, shortage of resources, demographic growth, change in society’s values, market liberalisa- tion and etc. Changes create new rules such as competition, energy efficiency, new business models, and energy production technologies. Regulations create new demand which raises questions: What can distinguish the company from others with new services or new business models? What can help to identify the potential of cost reduction? What is the most important information? How can I acquire this information quick and comprehensive? Existing processes and systems are not able to deliver the required answer in time, within budget and with a satisfactory level of quality. In the recent years, computational models are widely used in various fields and their use is growing in the energy industry too. More and more companies started to apply data mining methods for solving very wide range of tasks. For example, condition monitoring of high volt- age electrical equipment. Other example is data clustering techniques – such as the self-organ- izing map, that have been applied to vibration monitoring and analysis of transformer on-load tap-changers. Theoretical part of this paper briefly overviews new data mining methodology, inductive fuzzy classification. Practical part represents how this methodology is used for solving the problem of finding the boundary conditions of transformer station clusters. The purpose of the current case study is to apply theoretical knowledge in handling the practical challenge