CONTROL TECHNIQUES FOR IMPROVED PERFORMANCE OF MICROGRID INTEGRATED WITH RES

Abstract

Environmental, economic, and technological concerns have prompted the development of electrical systems based on the distributed generation (DG) model, which is associated with small-scale power generation and is primarily comprised of renewable energy sources (RES). These RES have significantly contributed to the advancement of microgrids, making them a viable alternative to conventional grids. Among all RES, solar photovoltaic (PV) is one of the most widely utilized due to its accessibility, ease of installation, and low maintenance requirements. Solar energy plays a significant role in environmental conservation and fostering a cleaner society. Additionally, advancements in technology have made PV modules increasingly cost- effective and efficient. However, due to their dependence on meteorological conditions for energy generation, solar photovoltaic systems (SPS) exhibit uncertainty in power output. Solar cells display nonlinear I-V and P-V characteristics that are influenced by external variables such as solar irradiation, humidity, temperature, geographical location, and various dynamic conditions. Consequently, the development of sophisticated control strategies is critical to ensure the efficient operation of solar PV systems. To address these challenges, a novel MPPT algorithm is developed by integrating the Incremental Conductance (INC) method with a double closed-loop voltage control strategy. This hybrid approach enables accurate MPP tracking while simultaneously regulating the DC bus voltage, thus enhancing the reliability of standalone hybrid microgrids. The proposed strategy is supported by a bidirectional DC-DC converter that facilitates intelligent charge/discharge control of the BESS, maintaining DC voltage stability within state-of-charge (SOC) limits. In the subsequent phase, the research advances to the development of intelligent, data- driven MPPT algorithms for grid-connected PV systems. An Artificial Neural Network (ANN)-based MPPT controller is designed and trained using a diverse dataset that captures the spatiotemporal variations in solar irradiance and temperature. To enhance the adaptability and generalization capabilities of the ANN model, the Horned Lizard Optimization (HLO) algorithm—a recent bio-inspired metaheuristic technique—is employed to optimally tune the ANN’s internal parameters. The resulting ANN-HLO MPPT controller demonstrates superior tracking accuracy, faster convergence, and robust performance under rapidly changing irradiance conditions. Performance robustness is validated using a three-month real-time solar irradiance dataset obtained from NASA and NREL for two geographically distinct locations: Shahabad Daulatpur (Delhi) in northern India and Chikkaballapur (Karnataka) in southern India. These datasets enable realistic and comprehensive testing of the MPPT controller under dynamic conditions, including irradiance fluctuations, temperature changes, voltage sag/swell events, and nonlinear load disturbances. Compliance with the EN50530 MPPT efficiency standard is also established for both fast and slow- v changing irradiance scenarios. Performance benchmarking confirms higher tracking accuracy, reduced settling time, and enhanced energy yield in comparison to existing techniques. Additionally, sensitivity analysis substantiates the algorithm’s robustness across a wide range of operating conditions. Recognizing the crucial role of inverter control in grid-tied systems, the thesis proceeds with an exhaustive study of DC link voltage regulation techniques. While conventional PI controllers are widely used, they often struggle to maintain voltage stability under fluctuating irradiance and nonlinear load conditions. To overcome this limitation, metaheuristic optimization techniques such as Cuckoo Search Optimization (CSO) and Honey Badger Algorithm (HBA) are introduced to optimally tune the PI controller gains (Kp and Ki). These techniques minimize integral error criteria and improve dynamic system response. The proposed control scheme maintains unity power factor operation and ensures harmonic distortion remains within IEEE-519 limits, even under complex nonlinear loading scenarios. The thesis also includes performance evaluation using statistical metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Square Error (MSE), offering quantitative validation of the proposed control strategies. Comparative analysis of conventional PI, ANN-PI, CSO-PI, and HBA-PI controllers is carried out through MATLAB/Simulink simulations under varying scenarios, including sudden load changes and irradiance dips, demonstrating the superiority of the optimization-enhanced approaches. In addition to advanced control methodologies, the thesis addresses one of the most critical protection challenges in grid-connected PV systems: islanding detection. A voltage ripple-based islanding detection method is studied, which accurately identifies grid disconnection events by analysing characteristic disturbances in the DC link voltage waveform. This method ensures minimal non-detection zones (NDZ), rapid response time, and compliance with IEEE 1547 and IEC 62116 standards. Overall, this thesis contributes significantly to the domain of solar PV-based power systems by integrating classical control theories with cutting-edge artificial intelligence and optimization methods. The proposed control strategies deliver improved energy efficiency, enhanced system reliability, and resilient performance under diverse environmental and operational scenarios. The outcomes of this work have direct applications in the design and deployment of next-generation smart microgrids, aligning with the global pursuit of clean, sustainable energy solutions.