Maximizing efficiency of solar energy harvesting systems supplying a microgrid using an embedded system
MetadataShow full item record
Due to the high initial capital cost of photovoltaic (PV) panels and their low conversion efficiency, it is imperative to operate the PV system at its maximum power point (MPP). In this context, the goal in this thesis is to develop and improve the PV system, by contributing to the optimization of energy withdrawn from the PV panel using an embedded system. For this purpose, the model of the PV panel is first studied in accordance with the real behavior of the PV panel. The single diode model of the PV panel is first developed in Matlab environment to obtain an excellent correspondence to manufacturer’s published curves and represent the effects of insolation and temperature on the solar characteristics. Secondly, the small-signal model for DC-DC Boost converter, with/without parasitic elements, operating in continuous conduction mode (CCM)/discontinuous conduction mode (DCM) is developed to understand the basic features of the switching system and investigate the effects of parasitic elements and losses on the model accuracy, the efficiency and the dynamic performance of the system. Next, this work proposes a design of a feedback controller for a DC-DC Boost converter to yield a robust, closed-loop controller structure with stable static and dynamic characteristics over the whole operational range of the converter. Moreover, a study of different numerical integration methods is presented to enable the implementation of real-time models in embedded hardware platforms. The primary objective of the research proposed is to control the output voltage or current of the PV array to generate maximum possible power at a certain irradiance and temperature. This can be achieved by implementing the maximum power theorem for load matching using the relationship between input and output impedances. To achieve this objective, the popular MPPT Perturb-and-Observe (P&O) technique is used as a test bench. This is followed by the development of a novel intelligent method using a Kalman Filter as an alternative to provide an acceptable performance against both the noises and dynamic environmental changes. Due to the excellent estimation ability of the Kalman Filter in the dynamic system within a noisy environment, an accurate MPP can be predicted without any reduction of system dynamics. A PV test system with a 100 kW PV array and MPPT controller using the traditional P&O and KF algorithms was used. The PV panel output power was controlled by three methods: (a) without any MPPT (i.e. open loop), (b) with an MPPT using the P&O algorithm, and (c) with an MPPT using the proposed KF algorithm. The obtained results clearly highlight the superiority of the proposed method with a very high level of robustness, reliability and accuracy. Furthermore, a Matlab simulation model yields an efficiency of 99.38 % under the standard test condition (STC), which is almost 5 % higher than the conventional P&O method under the same conditions. Finally, a hardware implementation of the system is tested on an FPGA chip Altera Cyclone II EP2C20F484C7 to verify the efficiencies and tracking speeds in a real-time environment.