PERFORMANCE ENHANCEMENT OF GRID CONNECTED SOLAR PV SYSTEM

Abstract

Solar photovoltaic (PV) based electricity generating system is becoming increasingly popular for managing energy consumption of remote and isolated communities. Solar energy source is the most promising renewable energy source (RES) due to its clean and unbounded supply, low maintenance requirements, lesser mechanical components, no green house gas emission, and ability to be put in remote regions for power generation. The design, control, and analysis of a grid-connected photovoltaic system have been given in this thesis work. It makes three-phase load balancing, active and reactive distribution adjustment, and power quality enhancement. The proposed system is implemented in both the single-phase and three-phase distribution networks. Maximum Power Point Tracking (MPPT) technique has been implemented to extract maximum power from the PV. A voltage source converter (VSC) in the system provides power compensation (active and reactive), harmonics elimination, load leveling, and enhancement in the overall power quality of the system. An appropriate control algorithm is needed for the proper operation of power electronic converters in a grid connected solar PV system. The control techniques ensure synchronization of different voltages, estimate fundamental components of voltages and currents and regulation of DC-link voltage of converters. Different conventional approaches for estimating synchronizing signals have been described, and novel advanced techniques for the systems have been proposed. Frequency Locked Loops (FLLs) and Phase Locked Loops (PLLs) are being extensively employed to estimate synchronization signals and can compute the phase, amplitude, and frequency of the load current. These approaches have been implemented under a variety of operating situations, including fluctuations in solar intensity and harmonics in the loads for the grid connected solar PV system. Conventional algorithms such as fourth ordered generalized integrator (FOGI), and Reduced-Order Generalized Integrator (ROGI) based FLL have been used for the proposed system. Furthermore, a novel technique using All-Pass Filter-PLL based control techniques vii have been presented, which are adaptive and provide fast dynamic response without compromising steady-state performance. Further advanced techniques like Coati Optimization Algorithm tuned Fuzzified-Phase Locked Loop (COA Fuzzified-PLL) and single-phase complex band-pass filter-based frequency locked loop (1ϕ-CBF-FLL) has been developed, rejecting the DC-offset and have faster dynamic response for the system. A Complex-Coefficient Reduced-Order Generalised Integrator-based Frequency-Locked Loop (CC-ROGI-FLL) control technique used for harmonic reduction. A novel ensembled Deep Reinforcement Learning (EDRL) MPPT controller has been developed. The ensembled DRL MPPT controller leverages the power of deep learning and reinforcement learning techniques to optimize the MPPT process, enabling the mitigation of inter-harmonics and efficient power extraction with a constant reference DC voltage at the DC link. In the present work, three primary components of the power circuit of the system are the grid, loads at common point of interfacing (CPI), and solar energy conversion system (SECS). A boost converter, VSC, an interacting inductor, and a ripple filter make up the power circuit of SECS. A boost converter is the first stage and a VSC is the second. The PV array is linked to the input of the boost converter. MPPT is served by the first stage of the boost converter. The output of the boost converter is linked to the DC link of VSC. The second stage of VSC adds PV power to the grid and aids in improving the PQ of the distribution network. The proposed system and algorithms have been simulated and examined in the MATLAB/Simulink environment. The experimental results validate the effectiveness of the proposed control schemes in a standalone system.