This paper introduces a novel hybrid technique, called RERNN-SCSO, for a unified power quality conditioner (UPQC) in an electric vehicle (EV) charging station (EVCS). The RERNN-SCSO technique combines a recalling-enhanced recurrent neural network (RERNN) and sand cat swarm optimization (SCSO) to forecast and control converter signals. The UPQC system includes series-active power filters (Se-APF), shunt-active power filters (Sh-APF), and a DC-side capacitor. The primary goal is to enhance power quality, addressing issues such as voltage sags, voltage unbalance, harmonic currents, and voltage disturbances. The system is integrated into an EVCS, where EVs are connected in parallel to the UPQC DC-side using a double-layer microgrid (MC) and DC–DC converter. The performance of the RERNN-SCSO method is evaluated using MATLAB/Simulink, showing lower THD under voltage sag and swell conditions compared to existing methods.This paper introduces a novel hybrid technique, called RERNN-SCSO, for a unified power quality conditioner (UPQC) in an electric vehicle (EV) charging station (EVCS). The RERNN-SCSO technique combines a recalling-enhanced recurrent neural network (RERNN) and sand cat swarm optimization (SCSO) to forecast and control converter signals. The UPQC system includes series-active power filters (Se-APF), shunt-active power filters (Sh-APF), and a DC-side capacitor. The primary goal is to enhance power quality, addressing issues such as voltage sags, voltage unbalance, harmonic currents, and voltage disturbances. The system is integrated into an EVCS, where EVs are connected in parallel to the UPQC DC-side using a double-layer microgrid (MC) and DC–DC converter. The performance of the RERNN-SCSO method is evaluated using MATLAB/Simulink, showing lower THD under voltage sag and swell conditions compared to existing methods.