José Almeida

The need to promote clean, green energy practices has been increasing to create a more sustainable economy. Integrating energy resources, such as renewable generation and battery energy systems, to promote a green economy creates new challenges for the electrical system. One solution to mitigate these problems is to leverage residential consumers' local flexibility through demand response participation. This work proposes an automated tuning of a genetic algorithm-based scheduling program for home appliances using an F-race iterated approach. Different case studies are considered based on the demand response programs and PV generation for a day of optimization. These case studies serve as training instances for the automatic configuration process, enabling irace to obtain a sound, robust configuration for algorithms that work across multiple instances. Results show the best elite configuration given by the iterated racing approach for automatic algorithm configuration. Comparing this configuration with the default configuration for each case shows that the auto-tuned GA yields better results than the default GA, with around a 56% decrease in electricity costs, mainly when a flat tariff is used, and PV generation is available. Still, the lowest electricity bill costs were obtained considering a time-of-use program with PV generation for autoconsumption.

The growing complexity of energy systems under smart grid paradigms has highlighted the limitations of deterministic optimization models. To address this, metaheuristic algorithms have emerged as robust alternatives to solving energy resource management problems. This paper presents Web Meta-ERM, a novel, web-based platform designed to facilitate testing, comparison, and sharing of metaheuristic algorithms in smart grid contexts. It enables simulations on real-world case studies and allows researchers to upload and reuse algorithms and datasets. The platform promotes open science, reproducibility, and collaborative experimentation. Using a test case based on the IEEE 33-Bus system, we demonstrate how the platform enables benchmarking of different metaheuristics, including those in the natural computing paradigm (e.g., particle swarm optimization, differential evolution). Web Meta-ERM advances optimization research by reducing setup barriers and supporting comparative performance analysis, educational engagement, and dissemination of energy management models.