Load flow estimaton in electrical systems using artificial neural networks

This paper studies the possibility to apply the approximation power of the MLP ANN in developing an alternative method for computing the load flow when the configuration of the system does not change. If the bus admittance matrix remains the same, the bus voltages are affected only by the changes of the bus active and reactive loads, in a non-linear and predictable variation. If the network is trained with a sufficient amount of bus loading scenarios, for which the resulting bus voltages are known, then the MLP network is capable of finding with a satisfactory precision the bus voltages when providing to it new bus P and Q values. Tests are carried out on the IEEE 14 bus system.