Steady-State Error Compensation for Reinforcement Learning-Based Control of Power Electronic Systems

Data-driven approaches like reinforcement learning (RL) allow a model-free, self-adaptive controller design that enables fast and largely automatic development process with minimum human effort. While it was already shown in various power electronic applications the transient control behavior for complex systems can be sufficiently handled by RL, challenge of non-vanishing steady-state errors remains, which arises from usage policy approximations finite training times. This is crucial problem require accuracy, e.g., voltage grid-forming inverters or accurate current motor drives. To overcome this issue, an integral action state augmentation RL controllers introduced mimics integrating feedback does not any expert knowledge, leaving approach model free. Therefore, learns how to suppress deviations more effectively. The benefit developed method both reference tracking disturbance rejection validated two source inverter tasks targeting islanded microgrid as well traction drive applications. In comparison standard setup, suggested extension allows reduce error up 52% within considered validation scenarios.