Abstract
In order to improve the typical voltage quality problems in distribution power system with the dynamic voltage restorer (DVR), some works are conducted in this paper. Firstly, according to the DC side voltage fluctuation of the inverter in DVR, the DC voltage is treated as a state variable and detected in real-time by the feed-forward loop. So the gain of inverter bridge is compensated for and the influence of the fluctuation of the DC input voltage on the compensatory output of the DVR system is offset. Secondly, for the deficiency of the traditional proportional integral (PI) control strategy in the DVR, a genetic PI algorithm combining with repetitive control algorithm is proposed. Theoretical analysis testifies that this compound control algorithm can effectively improve the dynamic and static performance of the system output. Lastly, the proposed control algorithm is comparatively validated through the extensive simulation studies, and the results show that the proposed compound control method can improve the response speed of the DVR system and compensation accuracy, greatly reduce the harmonic contents of the load voltage, and effectively improve the voltage quality of the sensitive load.
Introduction: The Challenge of Voltage Quality
Modern power grids are increasingly populated with nonlinear loads, such as high-efficiency motor drives and power supplies, which cause voltage sags, swells, and harmonic distortion. These fluctuations can be catastrophic for "sensitive loads"—including data centers, medical equipment, or precision manufacturing—leading to equipment failure or data loss.
The Dynamic Voltage Restorer (DVR) is a custom power device connected in series between the source and the load. When a voltage disturbance occurs, the DVR injects a compensatory voltage to maintain a stable load profile. However, traditional Proportional-Integral (PI) controllers often struggle with the nonlinear nature of DVRs and fail to eliminate periodic harmonic errors effectively. This paper introduces a high-performance control strategy to bridge this gap, creating a controller that is both fast (dynamic) and precise (static).
Proposed Compound Control Methodology
The authors propose a "Compound Control" architecture consisting of three main pillars designed to optimize the DVR's performance:
- Genetic PI Control: Traditional PI controllers use fixed gains ($K_p$ and $K_i$). The authors utilized a Genetic Algorithm (GA)—an optimization technique inspired by natural selection—to find the optimal gains. By minimizing the "Integral of Time multiplied Absolute Error" (ITAE), the GA ensures the controller responds aggressively to changes while minimizing long-term oscillation.
- Repetitive Control (RC): Based on the "Internal Model Principle," RC is designed specifically to eliminate periodic errors. Since harmonics in a 50Hz system repeat every 20ms, the RC "learns" from previous cycles to cancel out recurring distortions.
- DC-Link Feed-forward Compensation: Inverters in a DVR rely on a DC voltage source. If this DC source fluctuates, it affects the output. The authors added a feed-forward loop that detects DC fluctuations in real-time and adjusts the inverter gain to offset the influence before it reaches the load.
Simulation and Performance Analysis
As this is a control theory and power electronics study, the authors conducted extensive simulation studies in a MATLAB/Simulink environment. They modeled a single-phase low-voltage system and subjected it to various programmed disturbances, including 60% voltage sags/swells and injected 5th through 8th-order harmonics. A "sensitive load" model consisting of a diode rectifier and a DC/DC converter was used to test real-world applicability.
Key Insights
- The Hybrid Advantage: The core innovation is the synergy between PI and RC. PI control acts immediately at the switching frequency level but lacks precision regarding harmonics. Repetitive control is highly precise for harmonics but has an inherent one-cycle delay (20ms). Parallel operation allows the PI to handle sudden sags while the RC cleans up harmonic ripples.
- Dynamic DC Compensation: Treating DC voltage as a fluctuating state variable rather than a constant allows for a more robust system that handles internal hardware imperfections.
- Automated Optimization: Using a Genetic Algorithm removes the trial-and-error typically associated with tuning controllers for nonlinear power systems, leading to a more mathematically grounded performance boost.
Results and Conclusion
The simulation results demonstrated a dramatic improvement over traditional PI control:
- Harmonic Suppression: In a standard 60% variation test, the Total Harmonic Distortion (THD) dropped from 5.17% with traditional PI to just 0.59% with the compound method.
- Error Reduction: The peak error in load voltage was reduced from 0.13 pu to 0.031 pu (per-unit).
- Complex Scenarios: Under intentional harmonic injection, the compound controller maintained a THD of 1.65%, whereas the traditional controller spiked to 10.73%, which is well outside acceptable standards.
In conclusion, the proposed compound control method significantly improves the response speed and compensation accuracy of DVR systems. While the current study is simulation-based, it provides a strong theoretical foundation for future hardware implementation in industrial distribution networks. Future work may involve testing on physical DVR prototypes and expanding the logic to three-phase systems.
Summary updated