基于重复扩张状态观测器的自抗扰控制及并网逆变器应用

Grid-connected inverters face persistent challenges in current control due to voltage harmonics from the grid and nonlinearities such as inverter dead-time effects. These disturbances introduce low-frequency harmonics into the grid current, threatening power quality and system stability. Additionally, uncertainties caused by component aging and parameter drift in filter circuits further compromise control performance and may lead to instability. Active disturbance rejection control (ADRC) is a robust control strategy, leveraging an extended state observer (ESO) to estimate and compensate for disturbances and model uncertainties. However, conventional ESOs are limited in bandwidth and struggle to observe high-frequency harmonic disturbances typical in grid-connected inverters. As a result, the grid current exhibits high total harmonic distortion (THD) and significant steady-state error. Repetitive control provides strong harmonic suppression. However, since its compensator design relies on accurate modeling of the plant, system instability may occur in the presence of uncertainties. This paper proposes a novel repetitive extended state observer-based ADRC (RESO-ADRC). The disturbance estimation capability of the ESO is enhanced by integrating repetitive control into its structure. Both periodic disturbances and system uncertainties can be accurately estimated by the resulting RESO, allowing effective rejection of voltage harmonics and compensation for system uncertainties. The limitations of the conventional ESO are analyzed from the perspective of the internal model principle. Accordingly, repetitive control is integrated into the ESO, enhancing its disturbance estimation capability. A compensator is then designed based on the RESO rather than the controlled plant. To limit the observer bandwidth and reduce noise sensitivity, a zero-phase low-pass filter is introduced into the RESO, effectively reducing the system’s sensitivity to high-frequency noise. The paper also presents a discrete-domain stability analysis of RESO and RESO-ADRC. The tracking and disturbance rejection performance are analyzed, and optimized controller parameters are provided. Furthermore, robustness analysis under input gain mismatch is discussed. By switching between different LCL filters, the robustness and steady-state performance of the traditional ADRC, repetitive control, and the proposed RESO-ADRC are evaluated. The results show that the traditional ADRC maintains stability across different filters but exhibits poor steady-state performance. Repetitive control offers good steady-state accuracy but becomes unstable when the filter changes. In contrast, RESO-ADRC consistently achieves low THD and demonstrates robust performance across various filter configurations. In addition, the influence of input gain on the system's dynamic performance is investigated, and the stability margins and robustness conditions of the controller parameters are verified.