Abstract
Renewable energy injected into the utility grid is gaining increased attention in efforts to keep up with energy demands in the near future. The extensive integration of renewable power can reduce the usage of fossil fuels and environmental pollution. In particular, solar photovoltaic (PV) energy is growing exponentially due to its benefits, such as its being environmentally friendly, uses sunlight, with low maintenance costs, and reduced cost of utility bills. However, power quality and low efficiency are a concern in transmission and distribution networks, due to partial shading conditions (PSC), and emerging current and voltage harmonics; hence the need for maximum power point tracking (MPPT) algorithms, to continuously harvest global maximum power peak (GMPP) under complex weather conditions. Along with the power quality related-issues, comes the novel control techniques for controlling the grid-tied PV interfacing inverter to continuously mitigate harmonics, and enhance the quality of power, and system stability.
In this thesis, the first contribution provides a detailed design of a grid-integrated PV system by studying the Pretoria geographical location and real-time meteorological data, so as to optimize the energy yield of the PV modules. The study is aimed to evaluate the performance between monofacial and bifacial PV modules in both fixed-tilt and horizontal single-axis tracking (HSAT) configurations, using PVsyst simulation software. The purpose of this case study is to determine whether the proposed location is best suited for solar energy harvesting and to evaluate the superior technology between the monofacial and bifacial technologies. Furthermore, the optimal installation configuration is validated between fixed tilt and HSAT configuration. The high-performing PV module technology and the best installation configuration are then used in the next case studies. The simulation results demonstrate that the Pretoria location is best suited for solar renewable energies, as it can produce up to 4246 𝑘𝑊ℎ/𝑘𝑊𝑝 per year and expected annual global horizontal irradiance (GHI) of 2043.3 𝑘𝑊ℎ/𝑚2. The energy available at the inverter output and power injected into the grid is 4256 kWh/year. Additionally, the combination of bifacial PV modules and HSAT configuration demonstrated the best results, which are regarded as the lowest Levelized cost of energy, due to its reduced energy payback time.
The second contribution introduces a new hybrid MPPT algorithms with centralized assertive reinitialize conditions for GMPP tracking; and to maintain a stable DC voltage level at the
iii
common DC bus under PSC. The condition to restart the GMPP search is not widely discussed. The major challenge facing global search algorithms is related to restart conditions. Avoiding unnecessary searching and providing an assertive GMPP restart condition is critical for PV system operation. The proposed hybrid MPPT techniques utilized for this comparative study are: (i) Modified robust linear regression optimized by ant colony optimization (MRLR-ACO), (ii) Ensemble RUSBoosted tree – Linear programming (ERBT-LP), (iii) Stepwise linear regression optimized by Genetic algorithm (SLR-GA), (iv) Flexible radial movement optimization based on dynamic safety perimeter (FRMO-DSP), and (v) Deep belief network - Jellyfish search optimization (DBN-JSO). It was found that hybridization of algorithms improved the performance of the schemes and demonstrated the best results in all the case studies. The techniques extracted GMPP and minimized power losses with a neglectable oscillations around MPP. It is worth mentioning that the algorithms have a rapid convergence and tracking speed under any type of PSC. The algorithms are found satisfactory to be utilized to maintain the DC voltage at the common DC bus on a grid tied PV system.
The third contribution is twofold. The first part introduces novel control algorithms for grid-interfacing two level inverter control for harmonics mitigation. The proposed inverter control strategies are used to determine the three-phase reference voltage control signals to generate the optimum pulse width modulation (PWM) signals for grid interfacing inverter to improve the power quality in a grid-connected PV system. The proposed adaptive inverter control schemes proposed are: (i) novel inverter dual loop control system (IDLCS), (ii) Gaussian kernel trained Support vector machine (GK-SVM), (iii) hybrid Expectation-Maximization (EM) algorithm, (iv) Quasi-recurrent neural network (QRNN). The second part introduces the enhanced dual second order generalized integrator phase-locked loop (EDSOGI-PLL), for improving grid synchronization, harmonic attenuation, and symmetrical components estimation under steady-state and transient states. The proposed integrated system measures the total harmonic distortion (THD). Results suggest that the proposed algorithms comply with the IEEE 519 and IEC 61000-4-30 Standards to protect the system network from a high level of financial loss. It is worth mentioning that the proposed control strategies have benefits such as superior harmonic attenuation ability, enhanced stability, control adaptivity, adaptive frequency, and low computation burden.
Keywords: Photovoltaic, partial shading conditions, hybrid systems, machine learning, optimization, maximum power point tracking, power quality, total harmonic distortion.