Abstract: Aiming at the complex beam commissioning task of a 0.3 MV compact accelerator mass spectrometer (AMS), which is time-consuming and hard to optimize manually, a novel auto-tuning method combining the experimental physics and industrial control system (EPICS) and the differential evolution (DE) algorithm is proposed. An intelligent optimization algorithm core module is developed in Python and connected to the EPICS control framework via PyEpics for efficient data interaction, forming a complete automatic beam commissioning system. The system automatically tunes the power supply voltage and current, thereby adjusting the key parameters such as magnetic field strength, electrode voltage, and beam current. The DE algorithm dynamically adjusts population parameters for space search guided by real-time beam current feedback under equipment safety constraints. Experiments show that this method reduces the average convergence time from 2.5 h (manual tuning) to 30 min, increases the optimization success rate from 60% (traditional methods) to over 90%, and stabilizes the optimal beam current at 95% of the theoretical maximum. This approach significantly shortens the tuning time and enhances the stability and efficiency of the accelerator, offering substantial industrial value.