Congestion management in a zonal market by a neural network approach

With the introduction of the Power Exchange, one of the most critical issues to be faced by a Transmission System Operator (TSO) is to take into account the transmission constraints in a simplified market model. The zonal approach represents a suitable solution, since its mechanism can be easily understood by all the operators; on the other side, it requires to establish a priori the relevant transmission constraints. However, in a meshed network, this solution results in some problems in the management of the system, mainly because the Transmission Capability (TTC) value is deeply influenced by both demand and generation patterns. In order to face this problem, coupling the clearing process with an on-line TTC evaluation tool would represent the best solution, allowing the full exploitation of the transmission facilities. Since all the methods already proposed in the technical literature are not suitable for on-line applications due to their huge computation time, a new approach is proposed. An Artificial Neural Network (ANN) is used to estimate the TTC in real time: once the proposed model has been trained, it is adopted for a real time update of the TTC between two market areas, with respect to the actual market results, in order to increase the market efficiency and to reduce the associated congestion costs.