Abstract
Maintaining frequency stability is one of the biggest challenges facing future power systems, due to the increasing penetration levels of inverter-based renewable resources. This investigation experimentally validates the frequency provision capabilities of a real Polymer Electrolyte Membrane (PEM) hydrogen electrolyser (HE) using a power hardware-in-the-loop (PHIL) setup. The PHIL consists of a custom 3-level interleaved buck converter and a hardware platform for real-time control of the converter and conducting grid simulation, associated with the modelling of the future Iberian Peninsula (IP) and Continental Europe (CE) systems. The investigation had the aim of validating earlier simulation work and testing new responses from the electrolyser when providing different frequency services at different provision volumes. The experimental results corroborate earlier simulation results and capture extra electrolyser dynamics as the double-layer capacitance effect, which was absent in the simulations. Frequency Containment Reserve (FCR) and Fast Frequency Response (FFR) were provided successfully from the HE at different provision percentages, enhancing the nadir and the rate of change of frequency (RoCoF) in the power system when facing a large disturbance compared to conventional support only. The results verify that HE can surely contribute to frequency services, paving the way for future grid support studies beyond simulations.

Abstract
In a high-risk sector, such as power system, transparency and interpretability are key principles for effectively deploying artificial intelligence (AI) in control rooms. Therefore, this paper proposes a novel methodology, the evolving symbolic model (ESM), which is dedicated to generating highly interpretable data-driven models for dynamic security assessment (DSA), namely in system security classification (SC) and the definition of preventive control actions. The ESM uses simulated annealing for a data-driven evolution of a symbolic model template, enabling different cooperative learning schemes between humans and AI. The Madeira Island power system is used to validate the application of the ESM for DSA. The results show that the ESM has a classification accuracy comparable to pruned decision trees (DTs) while boasting higher global inter-pretability. Moreover, the ESM outperforms an operator-defined expert system and an artificial neural network in defining preventive control actions.