To Approach Minimum Losses of the Distribution Systems by Artificial Neural Networks

Loss reduction of distribution system can perform the switching operation to make feeder reconfiguration. First, we get the feeder hourly loading by considering the weather information and feeder historical load data or by SCADA system. Second, we determine current flow of each feeder section according to the load composition of residential, commercial and industrial customers. Third, by using the unsupervisory self-organize mapping network (SOM) to classify the section current combination, the network can not through training process. Finally, the supervisory neural network of back-propagation (BP) is used to derive the optimal feeder reconfiguration to reduce feeder loss, and the network must be through by training process till the network converged. With the proposed neural network, it can be applied to the real time system, to reduce the feeder loss of power distribution system by quickly and efficiently to get the switching status