Commande robuste de systèmes photovoltaïques interconnectés au réseau avec convertisseurs multi niveaux

Abstract

Several grid-connected PV topologies are investigated in this dissertation for low, medium and high power applications. Furthermore, an effective controllers have been proposed for these topologies instead of conventional control strategies. The latter’s suffer from several drawbacks such as bad MPPT tracking, inaccurate grid power control, low grid current quality, hard implementation practically and variable switching frequency. To overcome these drawbacks, a model predictive control (MPC) strategy is proposed to control the power converters employed in the investigated grid-connected PV topologies. The major drawbacks of MPC strategy are variable inherent switching frequency and computational burden especially in case of high-level inverters. Therefore, these drawbacks are taking into account in the design of proposed controllers. Firstly, an improved control strategy based on fixed switching predictive control strategy for three-phase dual-stage gridconnected PV system is proposed. A variable incremental current step size of an MPPT current oriented loop based on fixed-switching predictive current control is proposed and employed to control the DC-DC converter. While, a modified VOC based on predictive control strategy and space vector modulation (SVM) is employed to control the DC-AC converter. Afterwards, High-level NPC inverters are employed in grid-connected PV system in order to inject a high produced PV power into the grid with high performance operation. On the other hand, a simple and effective model predictive control (MPC) algorithm is proposed for grid-connected PV system using high-level NPC inverter (sixlevel) that permits to inject the active power generated by the PV system, the reactive power demanded by the grid operator and assure the balance of DC-link capacitor voltages. Then, an optimized model predictive control (O-MPC) is proposed in order to achieve the same performance control provided by the MPC algorithm but with a significant reduction in computational burden. Finally, a topology that divided the large PV array to string modules with individual DC-DC converter connected to a centralized multilevel inverter is investigated in order to overcome the problems resulted by connecting the PV modules as large PV array. Simple and effective controllers for this topology based on finite control set model predictive control (FCS-MPC) strategy is proposed. A voltage oriented maximum power point tracking (VO-MPPT) performed by an FCS-MPCC is applied for each DC-DC converter to draw the maximum power point from each string PV modules. In addition, an FCS-MPC controller is proposed to control the centralized multilevel NPC inverter connected to the grid. The simulation and HIL results validate the proposed control schemes for the investigated grid-connected PV topologies.