Cecilio Angulo

Decision-making in precision agriculture requires high-resolution information capable of representing interacting thermal and water-related constraints. This study introduces an interpretable geospatial artificial intelligence workflow that transforms daily meteorological station observations into structural agroclimatic zones supporting crop-suitability screening. Daily observations from 1, 090 stations operated by the Spanish State Meteorological Agency (AEMET) and the Automatic Weather Station Network of Catalonia (XEMA) are processed into 28 variables over ten agricultural campaigns, from November 2015 to October 2025. The features are interpolated onto a 500 m grid using elevation-assisted regression-kriging and organized into four families: thermal regime and accumulation, winter chill and frost exposure, hydroclimatic and water-balance indicators, and extreme thermal stress events. Sampled station-wise leave-one-out validation comprising 56, 000 predictions yielded median values of 0.782 for R2 , 0.888 for Pearson correlation and 0.869 for Spearman correlation. Regression-kriging improved both R2 and root mean square error (RMSE) relative to an elevation-only baseline for all 28 features. Principal Component Analysis retained eight components explaining 95.401% of total variance. A six-zone K-means solution was selected as a compromise between compactness, separation, spatial coherence and agroclimatic interpretability. A post-hoc Random Forest surrogate reproduced the cluster labels with 98.95% out-of-bag accuracy. Aggregated feature importance attributed 48.76% of cluster discrimination to thermal regime and accumulation, 31.66% to winter chill and frost exposure, 14.08% to hydroclimatic and water-balance indicators and 5.51% to extreme thermal stress. The workflow provides a scalable climatic baseline for precision agriculture, regional planning and climate-adaptation applications.

Sovereign agricultural data spaces face a coordination challenge: heterogeneous entities holding commercially sensitive agronomic data must be made interoperable and reusable by AI services without requiring per-entity bespoke adapters or sacrificing data sovereignty. We present the UPC Data Space Agro, a production sovereign data space with 187 contributing entities and 6.24 million RDF triples, built on the Agrixel semantic standard. The data space delivers six services organised around a shared semantic input layer: a Standardisation Service that converts heterogeneous entity data into Agrixel-standard files; a Standardised File Management Service for visualising and preparing standardised data; three AI services covering Soil Moisture Prediction, CO2/NEE Carbon Estimation, and Agroclimatic Clustering; and a Satellite Data Download Service for Earth Observation data retrieval. The shared ontology acts as a coordination protocol: services can be developed against a common representation and applied to contributing entities once their data have been standardised, while new entities can access the deployed services without requiring service-specific integrations.

Machine learning in healthcare struggles for one main reason: good data is hard to come by. Medical records are sensitive, and the rules on sharing them between hospitals are strict. A common workaround is synthetic data, that is artificial records that copy the statistics of real ones; however, on its own, it offers no real privacy guarantee, and earlier studies show it can still leak details about the patients behind it. We introduce an approach tackling both problems at once. It builds on a previous UMAP-based generator methodology and extends it so several hospitals can work together without sharing raw records, in a federated learning form. Then, the proposed methodology adds differential privacy so every shared quantity carries a formal (ε_DP,δ) guarantee, tracked by a custom privacy accountant. Two versions of the methodology are tested: a partially synthetic one, where a reference hospital validates the others’ rows, and a fully synthetic one, where each centre builds its own data from aggregated cluster statistics. Both are evaluated on four publicly available healthcare tabular datasets spanning prostate cancer (PI-CAI, CIA), breast cancer (BC-MLR), and cardiovascular disease (fiv-CardioDB), to test whether the observed trends generalize beyond a single clinical domain. As a key takeaway, it is demonstrated that going federated barely affects quality, but the privacy cost depends strongly on the moment in the procedure when the noise is injected, and this pattern holds consistently across all four datasets.

Modern power grids with high renewable energy penetration are vulnerable to fast voltage disturbances caused by grid faults. Among these, voltage sags are critical because they develop within milliseconds and require rapid reactive current support to maintain grid stability and power reliability. Reinforcement learning has previously shown potential for reactive current injection control during voltage sag events due to its fast response and adaptability to changing system conditions. However, existing approaches rely on separate policies for specific subsets of the operating space, which limits their ability to provide optimal actions when the system operates across broader or combined state regions. To address this limitation, this paper proposes a unified Soft Actor–Critic (SAC) target policy trained over the full state and action space by integrating multi-source transfer learning with potential-based reward shaping approach. Results show that the proposed multi-source transfer approach enables the target agent to converge faster and reach a higher reward solution than the baseline SAC and single-source transfer approach. The trained policy also improved prediction accuracy, achieving reactive-current errors below 0.2 A with respect to the ground-truth reference generated through extensive simulations over the full observation and action space. The reference follows the grid-code requirement for minimum reactive current injection during faults and provides a benchmark for evaluating prediction accuracy. This can help distributed generation sources respond more effectively during severe perturbations such as voltage sags, support voltage recovery, and reduce the risk of cascaded disconnections that could lead to unwanted blackouts. Additionally, the inference execution time is also sufficiently fast to satisfy the response-time requirement of voltage sag events, confirming the real-time feasibility of the proposed controller.