基于深度确定性策略梯度的光伏MPPT优化算法:局部阴影条件下的全局搜索

The integration of photovoltaic (PV) systems with the grid connected four-leg voltage source inverters (4LVSI) offers more efficient power conversion and distribution. However, the complexity of these higher-order systems presents significant control challenges due to the sensitive nature of both DC bus voltage and 4LVSI output currents, which introduce complex dynamic interactions within the 4LVSI control system and limit its overall control performance, particularly under system uncertainties and disturbances such as variations in irradiance, PV cell temperature, and grid voltage. To address these challenges, an enhanced active disturbance rejection control (EADRC) scheme based on a super-twisting sliding mode observer (STSMO) is designed for the outer DC bus voltage control loop. The STSMO is incorporated in the ADRC method for its fast uncertainties and external disturbances estimations, high resilience against uncertainties, and good immunity to measurement noise. The proposed STSMO-based ADRC approach effectively estimates and compensates for system uncertainties and external disturbances, thereby enhancing DC bus voltage dynamics and stability, improving resilience against uncertainties, increasing immunity to measurement noise, ensuring better steady-state accuracy of the DC bus voltage, and increasing system reliability while reducing cost and size. The effectiveness and superiority of the proposed control method for the PV grid-connected 4LVSI system are validated through both simulation and real-time studies using the OPAL-RT simulator. The results demonstrate robust performance under parameter uncertainties and disturbances, including variations in irradiance, PV cell temperature, DC bus capacitor value, and grid voltage sags.