Distribution Network Reconfiguration Based on Back Propagation neural networks

Network reconfiguration, which is of great significance to improve the nodal voltage profiles and reduce the active power losses, can be considered as a pattern recognition problem. A pattern construction method based on structural simplification of distribution network is proposed, where the distribution network structure is simplified by branch grouping and pattern of network structure. MILP model is used to describe the network reconfiguration patterns, which is resolved by Harmony Search Algorithm (HSA). The network reconfiguration patterns can be carried out by BP neural networks when the loads data is collected. 33-node system have been introduced to prove the efficiency of the proposed method.