Enhanced Voltage Regulation of PV-Interfaced Power Converters With Current-Sensorless Control Techniques

Abstract

Maximum power point tracking (MPPT) is generally considered necessary for photovoltaic (PV) systems aimed at maximizing the overall energy harvest, since disturbances caused by unpredictable weather conditions are ubiquitous. The PV array is characterized by a nonlinear behavior that changes significantly with the operating point and environmental variables, e.g., irradiance level, temperature, and shading conditions. The nonlinear feature can cause high variations in control dynamics and lead to compromised MPPT performance. Moreover, due to changing environmental conditions, PV-interfaced converters inevitably switch between continuous conduction mode (CCM) and discontinuous conduction mode (DCM). Two modes present significantly different characteristics. Therefore, finding a simple yet effective controller for such PV systems can be challenging. A new control scheme in the voltage-mode control (VMC) framework is proposed in this dissertation, which enables the mixed conduction mode (MCM) operation with consistently fast and stable performance. A single control law is made possible for the entire operation range, owning to the current-sensorless active damping technique. The effectiveness is validated with the theoretical analysis and experimental results. A settling time of less than 2 ms has been achieved upon abrupt transitions between 50% and 5% irradiance levels. The extended state observer (ESO) is promising for systems subject to disturbances and uncertainties. In this dissertation, the application of the ESO-based control scheme is expanded and applied to the PV systems. The development is based on an in-depth analysis of PV power systems regarding the nonlinear and time-variant features. Thanks to the disturbance rejection loop, the active damping function is naturally offered, which boosts the system damping without extra sensors or complex model-based tuning. The ESO-based control technique also significantly enhances the robustness against model inaccuracy. When the PV power interface is linked to the single-phase AC grid, another disturbance would be the double-line frequency 2fg ripple appearing at both the DC and PV links. The ripples limit the performance of extracting the maximum power from the PV generator. The controller shall be equipped with sufficient capability to reject the 2fg ripple at the PV link. With this purpose, a new ESO scheme, namely quasi proportional resonant extended state observer (QPRESO), is developed, enabling the precise tracking and rejection of fast-varying sinusoidal disturbance (FVSD) at the targeted frequency. It focuses on mitigating the steady-state disturbance originating from the DC to single-phase AC conversion and dynamic disturbances caused by environmental conditions such as solar irradiance and voltage reference changes. The advantages include simple and intuitive tuning, easy implementation, and enhanced transient and steady-state control performance. A 1.5 kW laboratory platform is constructed to verify the effectiveness and practicality of the proposed control scheme. A less than 1% steady-state ripple is obtained with a relatively low 1 kHz control bandwidth and 3 kHz observer bandwidth, which resolves the trade-off in the parameterization without using extensively high observer bandwidth. As discussed, optimal dynamic and steady-state performance is necessary to maximize the solar energy harvest. Alternative to QPRESO, this thesis recommends a novel adaptive mechanism preserving the advantages of two different observers, namely the classical extended state observer and recent reduced-order generalized proportional integral observer (roGPIO). Simulation and experimental results prove the improved performance regarding overshoot-free fast transient response and steady state against disturbances, even with small input filtering capacitance. Meanwhile, the observer switching happens seamlessly through the proposed mechanism, which is confirmed analytically and experimentally. Eventually, the robustness in responding to system uncertainties is dedicatedly discussed for PV systems in this thesis, which prompts the proposal of a new reduced-order ESO for improved robustness and further motivates the proposal of a combined ESO structure with enhanced tuning flexibility.