A neural network-based tool for preventive control of voltage stability in multi-area power systems

Abstract

In this paper a neural network-based tool is proposed to provide on-line preventive control strategies capable to restore a multi-area power system to a secure operating point when a voltage instability condition is going to be reached. These strategies are, respectively, based on reactive power control actions, generating nodes voltage–magnitude control actions and load curtailment actions. After decomposing the power system into an appropriate number of areas, for each area a feed-forward neural network is trained to give as output to a defined set of inputs an area-based voltage-stability index. The voltage stability of the whole power system is characterised by the minimum value among the area indices. Thus, for a given operating condition, the most critical area is identified and remedial measures are considered only for a restricted region of the system. Since each net receives `internal' inputs and inputs from the remaining areas, it is shown that using the resulting neural network architecture appropriate algorithms can be developed to carry out on-line a preventive control strategy based on the above-mentioned types of control actions. The effectiveness of the proposed approach is tested on the IEEE 118-busbar, 19-machine power system.