Advancing Sustainable Energy Solutions: AI Hybrid Renewable Energy Systems With Hybrid Optimization Algorithms And Multi-Objective Optimization In Portugal

Abstract

The effects of global warming are increasingly evident in our daily lives, with climate change, due to the evolution of the economy and the population, therefore it is necessary to take an active role in the development of solutions aimed at achieving a sustainable and carbon neutral future. The energy sector is one of the sectors of our society that contributes to global warming, with the use of coal, oil, natural gas, among others. This is therefore a sector that would benefit from the development of sustainable energy solutions, aimed at an energy transition focused on the use of natural resources. The use of renewable energy sources is, however, associated with the challenge of being intermittent, since it depends on the weather conditions, which is why it is necessary to combine two or more renewable energy sources to create a hybrid renewable energy system, to guarantee the system's reliability. This is the purpose of this dissertation, the optimization of a hybrid renewable energy system in the context of Portugal. The research began with a literature review in the field of artificial intelligence and energy, and nine studies were identified from the systematic literature review, which enabled the problem to be formulated. With this a hybrid renewable energy system integrating solar panels and wind turbines was designed and optimized using a hybrid algorithm. The hybrid algorithm developed integrates the Particle Swarm Optimization and Grey Wolf Optimization algorithms, with the aim of combining the advantages of each and thus approaching the global optimum. This algorithm was also implemented in parallel, to overcome the difficulties experienced in terms of execution time and use of computing resources. Three experiments were carried out on the hybrid algorithm developed to try to achieve the global minimum, however, it was only in the third experiment that it was possible to identify a local minimum with an ABF close to the global minimum. For this to be possible, experiment 3 integrated the changes mentioned in the two initial experiments as being necessary for a more efficient optimization.