Grey-box modeling for hot-spot temperature prediction of oil-immersed transformers in power distribution networks

Power transformers are one of the most costly assets in power grids. Due to increasing electricity demand and levels distributed generation, they more often loaded above their rated limits. Transformer ratings traditionally set as static limits, a controlled environment with conservative margins. Through dynamic transformer rating, rating is instead adapted actual working conditions transformers. This can help distribution system operators (DSOs) unlock unused capacity postpone grid investments. To this end, real-time information operating conditions, particular its hot-spot oil temperature, required. work proposes grey-box model that be used for online estimation forecasting temperature. It relies on limited non-intrusive measurements was developed using experimental data from DSO Jutland, Denmark. The thermal has proven able predict temperature high accuracy low computational time, which particularly relevant applications. With six-hour prediction horizon mean average error 0.4–0.6 °C. By choosing stochastic data-driven modeling approach we also provide intervals account uncertainty.