A new real-time approach for the load forecasting in the operation of sub-transmission systems

The Transmission System Operators (TSOs) need to know exactly the main real-time electrical network parameters in order to operate the power system in a secure and efficient way. This is made using field measurements and signals whose redundancy in the high voltage transmission system allows the State Estimation algorithm to verify the coherence of the available measurements and to detect possible bad data. Nevertheless, in the lower voltage sub-transmission systems, the needed redundancy is not available, due to different reasons. It is important therefore to find alternative solutions to estimate and define pseudo-measurements in the sub-transmission systems in order to have a coherent State Estimation. In this paper, a new approach for the load forecasting in the high voltage sub-transmission system is proposed and implemented, aiming at obtaining “pseudo-measurements”. The proposed approach starts from the load estimation through Multiple Linear Regression, Artificial Neural Network (ANN) and a combination of them. The methodology is used to learn the relationship among past variables related to the loads. The results are compared to find the most suitable technique for the TSO, with a compromise between complexity and reliability and demonstrated on a real 132 kV sub-transmission system operated by the Italian TSO.