Mehrdad Setayesh Nazar - Power and Water University of Technology Iran-Tehran

Optimal Restoration is one of the most important duties of electric distribution control centers. It is an essential task for distribution system management in normal and emergency conditions. Accuracy of this procedure is very essential to overcome the subsequences of device overload or abnormal voltages. However, because of limited time between on line calculation procedure and implementation of the optimization output results, speed of calculation is very important. To overcome this problem, a shift from on line calculations to off line calculations can be considered and a pattern recognition procedure for online operation based on off line database may be used. It has been shown that back propagation neural networks can achieve the pattern recognition goals for different conditions, but one of the most draw back of the mentioned method is their slow convergence. This may reduce the effectiveness of ANN, because many ANN must be trained for different topology and load conditions of system. A neural network architecture that does not suffer from the above-mentioned drawbacks is the Fuzzy ARTMAP (Adaptive Resonance Theory-supervised predictive Mapping) neural network. In this paper, implementation of the Fuzzy ARTMAP neural network for defined problem has been proposed. In our proposed algorithm, genetic algorithm optimization for optimal restoration has been used for off line stage. The case study, which is done for Hakimieh city-state of Tehran, shows an accurate result and it is compared with the best answers.

Distribution Automation, Optimal Restoration, Fuzzy ARTMAP Neural Network Decision Support System