Neural network aided fractional-order second-order sliding mode controller for frequency stabilization of wind integrated nonlinear power system.

In this study, we evaluated the effectiveness of a fractional-order second-order sliding mode controller (FOSOSMC) in regulating the frequency of interconnected power systems (IPS) with the presence of nonlinearities, including a generation rate constraint and governor dead-band. Two-area IPS with a wind turbine generator were used to assess the performance of the suggested controller. To improve the performance, an estimation based on a neural network observer (NNO) was made to predict the mixed system errors, such as unknown external disturbances, parameter uncertainty, and nonlinearities. The estimated output of the NN-based estimator was then used in an improved FOSOSMC that could better handle system disturbances. This paper proposes a neural-network-observer-assisted fractional-order second-order sliding mode controller (NNO-FOSOSMC) for load frequency control of a wind-integrated nonlinear power system Simulation results show that the proposed method reduces peak frequency deviation to − 0.018 Hz compared with − 0.031 Hz for NN-FOISMC and − 0.045 Hz for NN-ISMC, achieving an improvement of about 42%. The settling time is reduced from 18 to 10 s, while control chattering is decreased by approximately 55%. These results confirm superior robustness and dynamic performance under load disturbances, renewable intermittency, GRC, and governor dead-band nonlinearities.