Optimal coordinative design of SVC and PSS with the application of neural network to improve power system transient stability

This study presents an Artificial Neural Network (ANN)-based coordinated control approach that integrates Static VAR Compensators (SVC) and Power System Stabilizers (PSS) to enhance the transient stability of power systems. The proposed method is tested on a two-area, two-machine, three-bus system supplying a 5000 MW resistive load using MATLAB/Simulink. The system includes two generating plants rated at 1000 MVA and 5000 MVA, respectively. A feedforward ANN with two hidden layers (each containing five neurons) is trained using 60 % of the data and the Levenberg-Marquardt backpropagation algorithm. Simulations involve two main fault scenarios: a single line-to-ground (SLG) fault applied at 4 s and cleared at 4.2 s, and a three-phase-to-ground (LLL-G) fault cleared at 4.1 s. Without any controllers, the system shows significant instability, with rotor angle deviation exceeding 90° and voltage sags of up to 0.5 p.u. When PID-based Generic-PSS is applied, stabilization is observed after 6 s, with residual speed oscillations of ±0.02 p.u. However, the ANN-based Generic-PSS reduces recovery time to approximately 3.2 s, enhances damping by over 40 %, and decreases voltage overshoot by around 25 %. Furthermore, the ANN-based MB-PSS in combination with SVC confines bus voltage deviations to within ±0.01 p.u. and speed deviations below ±0.005 p.u., even during severe LLL-G faults. Overall, the ANN-based controllers outperform conventional PID-based approaches by providing faster damping, improved voltage regulation, and enhanced robustness under fault conditions, making them a promising solution for smart grid applications.