MULTI-OBJECTIVE OPTIMAL REACTIVE POWER DISPATCH FOR HYBRID ENERGY SYSTEMS

Abstract

Optimal Reactive Power Dispatch (ORPD) is an important aspect of modern power system operation that deals with the effective management and control of reactive power while satisfying nonlinear equality and inequality constraints. The increasing integration of renewable energy sources (RESs), coupled with the stochastic nature of load demand and generation, has significantly increased the complexity and uncertainty associated with ORPD, rendering conventional optimization approaches inadequate. Consequently, the development of robust, efficient, and scalable optimization frameworks has become imperative for addressing both deterministic and stochastic variants of the ORPD problem. This thesis presents a systematic and comprehensive investigation into advanced optimization methodologies for solving deterministic, stochastic, and multi-objective ORPD problems. Initially, a novel hybrid optimization algorithm is proposed by integrating Opposition-Based Learning (OBL), Differential Evolution (DE), and the Giza Pyramids Construction (GPC) algorithm. The proposed hybridization strategically enhances the exploration–exploitation trade-off by incorporating population diversity mechanisms and adaptive search strategies. The effectiveness of the proposed method is rigorously validated using standard unimodal and multimodal benchmark functions, followed by its application to deterministic ORPD problems on IEEE 30-bus and IEEE 57-bus systems. The obtained results demonstrate significant improvements in system performance, including reduced real power losses, improved voltage profiles, and enhanced voltage stability, compared to existing state-of-the-art metaheuristic algorithms. Furthermore, the framework is extended to stochastic ORPD by incorporating probabilistic representations of uncertainties associated with load demand, wind speed, solar irradiance, and small-hydro inflows, thereby capturing the inherent variability and intermittency of renewable resources. Subsequently, the proposed hybrid algorithm is extended within a multi objective optimization paradigm through its integration with the Non-dominated Sorting Genetic Algorithm II (NSGA-II). This formulation facilitates the si multaneous optimization of multiple conflicting objectives, resulting in the x generation of a well-defined Pareto-optimal solution set. The performance of the proposed multi-objective framework is evaluated using established benchmark test suites, including ZDT and DTLZ,andisquantitatively assessed through widely recognized performance indicators such as Inverted Generational Distance (IGD), Hypervolume (HV), Generational Distance (GD), Maximum Spread (MS), and Spacing (S). The results indicate superior convergence characteristics and diversity preservation capabilities, outperforming several contemporary multi-objective optimization techniques. The developed framework is further employed to solve both deterministic and stochastic multi-objective ORPD problems. In addition, a machine learning-assisted optimization framework, termed Improved Evolutionary Algorithm–Self Organized Maps (IEASOM), has been developed to further enhance the solution search process. The incorporation of Self-Organized Maps (SOM) enables adaptive learning and efficient mapping of the high-dimensional search space, thereby improving convergence behavior and solution robustness. The proposed IEASOM method is applied to deterministic and stochastic ORPD models using the IEEE 30-bus system. Its performance is further evaluated on large-scale power networks, including the IEEE 118-bus test system. The results confirm the scalability and computational efficiency of the proposed approach. Overall, the proposed methodologies provide a unified and robust framework for solving complex ORPD problems under both deterministic and uncertain environments. This research work contributes to the advancement of power systems optimization techniques. The proposed approaches support efficient power system operation and improve voltage security. They also assist in the effective integration of RESs into modern power networks.