Photovoltaic (PV) systems are one of the main actors in distributed power generation. In particular, in urban contexts, the PV generators can be subjected to mismatching phenomena due to the different orientation of the modules with respect to the sun rays or due to shadowing. In these cases, the maximum power point tracking (MPPT) function must be designed carefully. In this paper, architecture, including one dc/dc converter for each PV generator, is considered. The converters' output terminals are series connected to a high-voltage dc bus, where also a bidirectional dc/dc converter managing the power from/to a storage device is plugged. The functional constraints deriving from the dc/dc converters' connection, the mismatching phenomena, the MPPT capabilities of the inverter, connected with its input terminals at the dc bus, are taken into account in order to determine the best operating point of the system as a whole. The real-time constrained optimization problem is solved by using the particle swarm optimization method, which needs the knowledge of the actual current versus voltage curve of each PV generator. The practical impact of this need is also discussed in the paper. The feasibility and the performances of the proposed approach are experimentally validated by using a laboratory prototype.

Two-stage photovoltaic (PV) configurations have become increasingly popular due to the decoupling between the inverter dc-link voltage and the PV voltage, adding flexibility to extend the MPPT range. However, the additional dc-dc converter increases the power converter losses. The concept of partial power converters (PPC), which reduce the amount of power handled by the dc-stage can mitigate this effect. However, the type of topology, its power and voltage rating, efficiency, and operating range can vary significantly depending on the function (boosting or reducing voltage) and type of PV application and scale (micro-, string- or multi-sting inverter). This paper analyses of the possible configuration of connections of PPC depending on the application and scale of the PV system, and introduces a new buck-type PPC. Three solutions for practical PV systems are further elaborated, including experimental validation. Results show that the PPC concept greatly improves the overall PV system efficiency with the added benefit that the dc-dc stage power ratings achieved are only a fraction of the PV system, reducing size and cost of the power converter without affecting system performance.

For large-scale photovoltaic (PV) systems, the multistring configuration is becoming more and more attractive compared with the classical central inverter, since it results in better energy yield by realizing distributed maximum power point tracking. Among the existing solutions, an attractive topology consists in a single dc bus bar collector cascaded H-bridge (CHB) inverter. Through the use of a single dc bus bar collector, the CHB inverter presents inherent balanced operation while the multistring PV-system is fully decoupled from the grid-tie inverter. This paper proposes the experimental validation of this structure on a reduced-size single-phase laboratory prototype. Results confirm the interest of the proposed PV multistring architecture.

Online knowledge of dc-dc converters behavior is always of great interest. Based on a reliable model of the converters, some great improvement can be achieved. First, the control of the converters can be designed more precisely, especially while thinking in nonlinear theories model-based controls, and it can help to improve the energy management and the efficiency. Also, the knowledge of the converters gives some really useful indications about their state of health and, then, represents a good diagnosis tool and fault detection possibility. This paper proposes a modeling of the converters and a new state observer dedicated to an online estimation of the model parameters. The proposed average models include parameter modeling the losses and their estimation. They are validated on two different converters: the classical dc-dc boost converter and the current-fed dual-bridge dc-dc converter (CFDB-also called isolated boost). It is shown that the model of this last converter is strongly nonlinear, which impacts on the estimation. Simulations and experimental validation are given both on the boost and the isolated boost, and comparisons with the Luenberger state observer and extended Kalman filter are given to underline the interest of the proposed parameter estimation in terms of convergence for nonlinear systems and convergence rapidity.

Recently, hybrid electrical power sources composed of storage elements and renewable energy sources are known to have made great development. These energy sources are connected to a dc bus and need a dc-to-ac converter to transfer the produced energy to the grid. Three-leg voltage source inverters equipped with an output LC filter are often used. The main objective of this stage is to generate a three-phase sinusoidal voltage with defined amplitude and to ensure the smallest harmonic distortion rate of the output voltage for any load conditions. To satisfy the defined objectives, we present in this paper a new control method based on differential flatness control technique. The main interest of this control method is the possibility to define the behavior of the state variable system in the steady state as well as in transients. The use of only one control loop allows obtaining high dynamic properties of the system which ensure small harmonic distortion rate of the output voltage. Experimental results under balanced, unbalanced, and nonlinear load conditions are presented and validate the effectiveness of the proposed control methods.