Abstract

The integration of a renewable energy distributed generation into microgrids poses a significant constraint in the way power is managed, further so due to the natural variability in renewable generation and the variability in the load demands. To address these issues, this paper introduces a novel approach to the Spider Swarm Optimization (SSO) algorithm, the Dynamic LoadAdaptive Power Splitting (DLAPS) strategy, to enable real-time adaptive power sharing and enhance system resilience. Unlike the classical methods of power allocation that are static, according to which the power is divided between sources of renewable energy and storage systems, and between these sources and critical loads, the DLAPS-SSO applies the idea of a machine learning based predictive model to predict the power and dynamically optimize power allocation between the sources of renewable energy and storage systems and the sources and the critical loads. The model provides a multi-objective optimization framework that aims to minimize power losses and grid frequency variations, and to maximize the system’s resilience to disturbances, including disconnection from the grid, component malfunctions, and the availability of renewable energy sources. The comparison of simulation results with those of the Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) methods shows that the energy efficiency of the DLAPS-SSO increases by 15%–20%, and the amount of power losses across various load profiles decreases by 30%–35%. Moreover, the proposed solution offers 60% faster recovery time in case of grid disconnection, maintains 65.9% of the critical load in case of component failure, and provides 40%–50% less resilience than state-of-the-art techniques. The analysis of seasons and real data shows that there is stability of the behavior with the increase of efficiency (18%–22% during winter, and 23%–25% during summer), and the ability of the suggested approach to be robust when changing plant configuration/operation. Integration of optimization of dynamic load management and adaptive power splitting will spur microgrid control strategies and offer a viable strategy to stabilize the grid, reduce operation costs, and enable sustainable changes in energy transformations. The results demonstrate the essential role of bio-inspired optimization and reactivity in the next generation of smart grids.