The continuous increasing penetration of Distributed Generation systems (DGs) into Distribution Networks (DNs) puts in evidence the necessity to develop innovative control strategies capable to maximize DGs active power production. This paper focuses the attention upon this problem, developing an innovative decentralized voltage control approach aimed to allow DGs active power production maximization and to avoid DGs disconnection due to voltage limit infringements as much as possible. In particular, the work presents a local reactive/active power management control strategy based on Neural Networks (NNs), able to regulate voltage profiles at buses where DGs are connected, taking into account their capability curve constraints. The Neural Network controller is based on the Levenberg–Marquardt algorithm incorporated in the back-propagation learning algorithm used to train the NN. Simulations run on a real Medium Voltage (MV) Italian radial DN have been carried out to validate the proposed approach. The results prove the advantages that the flexibility of the proposed control strategy can have on voltage control performances, generation hosting capacity of the network and energy losses reduction.
Active management of renewable energy sources for maximizing power production
CALDERARO, Vito;GALDI, Vincenzo;MASSA, GIOVANNI;PICCOLO, Antonio
2014-01-01
Abstract
The continuous increasing penetration of Distributed Generation systems (DGs) into Distribution Networks (DNs) puts in evidence the necessity to develop innovative control strategies capable to maximize DGs active power production. This paper focuses the attention upon this problem, developing an innovative decentralized voltage control approach aimed to allow DGs active power production maximization and to avoid DGs disconnection due to voltage limit infringements as much as possible. In particular, the work presents a local reactive/active power management control strategy based on Neural Networks (NNs), able to regulate voltage profiles at buses where DGs are connected, taking into account their capability curve constraints. The Neural Network controller is based on the Levenberg–Marquardt algorithm incorporated in the back-propagation learning algorithm used to train the NN. Simulations run on a real Medium Voltage (MV) Italian radial DN have been carried out to validate the proposed approach. The results prove the advantages that the flexibility of the proposed control strategy can have on voltage control performances, generation hosting capacity of the network and energy losses reduction.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.