An artificial neural network for optimum topology in distribution expansion planning

An artificial neural network supported with an expert system is developed to determine the optimum topology in the expansion of distribution systems. This technique reduces considerably both training and testing time. The required input data is a digital simulation of the overloaded feeders switch status which is obtained directly from the expert system. The output of the neural network is the corresponding optimum topology. The neural network is implemented in MATLAB using backpropagation learning rules, while the expert system is designed using PROLOG. The developed neural net is tested on a distribution system. Test results are reported to show good performance