<p>This study improved the sliding mode control (SMC) technique based on radial basis function (RBF) neural network for three-phase uninterruptible power supply (UPS). The proposed control technique is based on the method of adjusting the power flow of the UPS inverter to reduce the THD (Total Harmonic Distortion), the CMV (Common-Mode Voltage) and ensure asymptotic stability. In fact, classical passive control methods will produce steady-state errors in the output voltage, which is caused by the absence of a current control loop inside the control input. Therefore, the proposed control scheme is developed by adding a current control loop to eliminate the steady-state errors of the output voltage and reduce the harmonics of the current in the main circuit without losing the stability of the system. For more details, the sliding surface is designed to include an outer control loop that controls the voltage so that the voltage on the load tracks the desired voltage. Next, an inner control loop controls the inverter current to the desired value. In addition, the sliding surface is adaptively adjusted to change over time by the RBF neural network to reduce the chattering phenomenon and enhance the robustness of the control system. Finally, the effectiveness of the proposed control strategy will be verified by simulation and experimental results considering sudden changes in load as well as desired voltage along with presence of noise.</p>