راه حل سریع شبکه های توزیع شعاعی با جبران خودکار و پیکر بندی دوباره

Optimal operation of radial distribution networks with automated compensation and reconfiguration requires the solution of a combinatorial optimisation problem, since the variables are the on/off status of capacitor banks and the open/close status of tie-switches. The solution approaches recently proposed use iterative algorithms such as genetic algorithms, simulated annealing and tabu search, for which the network needs to be solved in different configurations and at different compensation levels. The aim of this evaluation is that of attributing a quality index to each solution so that all the solutions can be suitably ordered. In an automated network, any configuration can be obtained from another one through a set of elementary moves, each consisting of the change of status of a pair of tie-switches. In the same way, any compensation level can be obtained from another one through a set of elementary moves, each consisting of the change of status of a capacitor bank. Once the most important quantities of the network in the starting configuration are calculated, those in the final configuration can be determined assessing the variations due to the series of elementary operations performed on the network that are necessary to go from the starting configuration to the final one. In the present paper, the expressions for the calculation of power losses and bus voltage variations in a radial network due to an elementary reconfiguration move or an elementary compensation move are developed. These expressions are obtained on the basis of the hypotheses usually valid for distribution networks. They can be easily integrated in any algorithm for optimal reconfiguration and compensation, making easier its implementation and faster the solution attainment. Simplified feeder models for distribution feeders with many loads, have also been developed in order to further reduce computation time. Substitution of all feeders with an equivalent one having only one ending load, together with performing evaluations only for a reduced number of network branches, gives rise to remarkable savings, in terms of computation efforts, growing with the network size and the number of loads.