An optimal neural network to design generators and stabilizers for multi-machine power systems based on a promoted firefly algorithm

The purpose of this article is to investigate power system stabilizing (PSS) in multi-machine power systems. In this study, special attention has been given to the role of generator and network modelling which has a direct impact on PSS design. For this purpose, the most important generator models in a power system with several machines in the power network without connection to the infinite bus, and the network connected to the infinite bus have been simulated, and the effects of these models and the infinite bus on the dynamic conditions of the system have been considered. The results of the presented models and the desired network in PSS design have been investigated. To achieve this purpose, an optimal artificial neural network has been utilized, where the parameters of the PID controller are the network output. The network has been optimized by a new promoted version of the firefly algorithm for PSS design and the parameters of this controller in a number of specific working conditions in a multi-machine power system. The method of the optimized neural networks (ANN) has been used for communication and effective use of the parameters obtained through the promoted version of firefly algorithm in a continuous and wide workspace. Numerical simulations considering three-phase short-circuit situations show that ANN/PFF-PSS can decrease load angle overshoot (35.7%) and settling time (28.6%) compared to the conventional PSS. The recovery of voltage is improved also by 9.3%. Through an analysis of systems with and without an infinite bus, the robustness of the proposed stabilizer is validated and shown to be preferred for damping inter-area as well as intra-area oscillations in complicated power networks.