Eliminating voltage violations in power systems using secondary voltage control and decentralized neural network

An electrical power system is a large-scale system for which voltage control is of a significant concern for the dispatchers. For the purpose of voltage control, such a system can be partitioned into smaller size control regions. This means adapting the traditional power system to a more reliable and smart system able to cope with random and sudden load variations and disturbances. This paper investigates a decentralized neural network secondary voltage control. A number of representative buses are labeled as pilot buses displaying the critical point for voltage control in each region. The control uses decentralized neural network controllers which are attached to these pilot buses to eliminate steady-state voltage violations. The methodology is applied to the IEEE 118-bus network. The results show the performance and ability of partitioning and neural network secondary voltage control to regulate the voltage and to avoid propagation of disturbances between regions.