Self-Tuning PI Control of Voltage Source Converters of Distributed Energy Resources

Renewable energy sources can be effective in improving the quality of distributed generation due to their characteristics like small size and adaptability. Uncertainties in renewable power generation have led to the inability of classical controllers to establish proper operation of the power system over a wide range of operating conditions. This research introduces a new method for improving the performance of the PI controller by analyzing the fuzzy critic evaluations and the concepts presented in the field of reinforcement learning. Since the critic agent operates based on fuzzy logic and linguistic evaluations, we will not encounter complicated mathematical calculations. As a result, in addition to the reduction of computational complexity and runtime of the controller, we add a level of intelligence and adaptability. The simulation results, demonstrate a noticeable improvement in the performance of these controllers in different operating conditions.