Luis Orihuela

This paper presents a multi-layer modelling methodology for simulating water dynamics in unsaturated soils, designed to maintain a conservative balance and reduce computational costs compared to continuous models based on Richards’ equation. The model incorporates infiltration, evaporation, capillary redistribution, transpiration and gravitational drainage using an explicit time integration scheme. Technical validation is carried out by comparison with HYDRUS-1D, simulating a complete 120-day strawberry growing cycle under identical forcing conditions and hydraulic parameters. The results show high dynamic agreement, achieving an RMSE=1.19mm and a reduction in computation time of over 97%. Subsequently, the methodology was experimentally evaluated in April 2026 at a greenhouse strawberry plantation in Moguer (Huelva), using actual soil moisture measurements at different depths. The results obtained demonstrate the model’s good predictive capability, achieving an RMSE=4.71%VWC and a Willmott index dW=0.833 for the cumulative profile.

Monitoring nitrogen species in the soil solution is essential to optimize agricultural fertilization, but their in situ measurement is costly and time-consuming. Ion-selective electrodes (ISEs) represent a cost-effective alternative, although they have limitations: they cannot detect all species, and they introduce noticeable errors in the ones they do measure. To overcome this, this work proposes an Extended Kalman Filter (EKF) aimed at the joint estimation of concentrations (states) and kinetic rates (parameters) of the nitrification process. The algorithm relies solely on noisy measurements of two states (ammonium and nitrate). Its validation on experimental data demonstrates that the estimator robustly reconstructs the system dynamics against different parameter initializations and the presentation order of the experiments. Furthermore, this methodology successfully infers the intermediate state of the process, for which empirical measurements are unavailable.

This work presents a cascaded control strategy for underactuated autonomous surface vehicles, integrating an extended state estimator within a path-following controller. Although path-following control has been extensively addressed in the literature, its experimental implementation continues to face significant challenges, particularly in real-world environments characterized by dynamic uncertainties, unmodeled disturbances, and measurement noise. The proposed architecture preserves the classical structure of cascade control, applying a surge-guided line-of-sight-based guidance law at the kinematic layer and a modified robust-adaptive backstepping controller at the dynamic layer. Both controllers take advantage of a nonlinear set-membership extended state observer, reformulated to incorporate a recently identified kinetic model that accounts for uncertainties in both the inertial dynamics and the propulsion system. The main contributions of this work include the reformulation of the observer to incorporate the identified model and the integration of angular velocity measurements, as well as the demonstration of the input-to-state stability of the low-level tracking errors with respect to estimation errors. Another significant contribution is the experimental validation of the complete strategy on a catamaran-type autonomous surface vehicle, operating under real-world conditions with unmeasured disturbances. The experimental results demonstrate robust and accurate performance in both estimation and control across various circuit exercises.

Battery level management is crucial for the reliable operation of sensor networks. In this paper, we present a novel framework that leverages graph theory and computational geometry to enhance battery management, optimize communication, and maintain connectivity. Our approach builds weighted Delaunay Graphs as combinatorial models for sensor networks and investigates theoretical aspects of their hamiltonicity. We introduce six algorithmic methods addressing two main objectives. The first set focuses on battery management by identifying low-battery sensors, computing minimum spanning trees and shortest paths and performing edge contraction to simplify network topology while preserving connectivity. The second set examines the hamiltonicity of Delaunay Graphs using graph labeling and combinatorial techniques to analyze the existence of Hamiltonian paths. Our methodology integrates key graph procedures such as minimum spanning trees, shortest path algorithms, edge contraction, and graph labeling with computational geometry tools like Voronoi diagrams and Delaunay triangulation. This combination allows for an accurate modeling of spatial relationships within the network and dynamic adaptation to battery constraints. A case study on monitoring a sugar cane field demonstrates the practical application of our methods. The computational study reveals the efficiency and robustness of the proposed algorithms in optimizing battery usage and improving network performance in real-world scenarios. In conclusion, our work not only addresses critical battery management challenges in sensor networks but also deepens the understanding of the interplay between graph theory and network optimization. This integrated approach paves the way for further research in both theoretical and applied domains, offering promising solutions for enhancing the sustainability and reliability of sensor networks

En este artículo se presenta una primera aproximación al problema de identificación no paramétrica de sistemas muestreados asíncronamente. La solución propuesta parte, en primer lugar, de la generación de un cuaderno de trayectorias, que se ajustan a partir de los datos asíncronos disponibles. Posteriormente, se presenta un mecanismo de aprendizaje no paramétrico basado en interpolación de Lipschitz. Éste permite realizar predicciones, con error acotado, del valor de los estados del sistema. El mecanismo de aprendizaje, aunque basado en la literatura, es novedoso, al tener que manejar trayectorias n-dimensionales, en vez de vectores de n dimensiones. Se valida el método sobre el sistema dinámico caótico conocido como el atractor de Lorenz.

En este trabajo se revisita un problema clásico: el diseño óptimo de actuadores y/o sensores para sistemas lineales e invariantes en el tiempo. Considerando el sistema descrito en el espacio de estados, el objetivo sería el de escoger, bajo ciertas restricciones, la matriz de entrada B y/o la matriz de salida C para optimizar algún criterio asociado, habitualmente, con el grado de controlabilidad y/u observabilidad del sistema. Este trabajo propone resolver este problema empleando algoritmos genéticos. La principal ventaja de esta herramienta es que permite introducir las restricciones de forma muy sencilla, permitiendo resolver el problema de una forma mucho más general de las que existen en la literatura. Además, esta metodología permite considerar optimización multi-objetivo de forma directa, pudiendo analizar la situación en la que más de uno de estos criterios sean interesantes para la aplicación.