Abstract— This paper proposes an enhanced fractional-order complementary terminal sliding mode (FOCTSM) control strategy for active power filter (APF), incorporating an improved reptile search algorithm (IRSA) with reverse learning strategy and the Cauchy mutation. The proposed method leverages the inherent advantages of fractional-order control, including superior design flexibility and precision, to address the critical requirements of harmonic suppression and current tracking in APF applications. Firstly, a FOCTSM surface is designed based on current error, followed by the development of a new exponential sliding mode reaching law. Secondly, IRSA is employed to optimize the fractional-order parameters and control gains. Finally, the simulation model of the APF is built by the SIMULINK toolbox, and the simulation verification and analysis comparison are conducted. The simulation results reveal that the current total harmonic distortion (THD) is reduced from 25.41% to 0.67% after filtering. Moreover, under disturbances, the current tracking absolute error remains below 0.2. These results demonstrate that the proposed optimal control strategy improves both harmonic suppression and robustness in the APF system.