This paper solves the buck-boost converter operation problem in the discontinuous conduction mode and the feeding a DC bus of a combined battery/solar-powered electric vehicle grid. Since the suns radiation has a very important effect on the performance of photovoltaic solar modules due to its continuous variation, the main task of the system under study is the regulation of the output voltage from an MPPT system located at the output of the panels in order to obtain a DC bus voltage that is fixed to 24 V. This is ensured via a double-loop scheme, where the current inner loop relies on sliding-mode control; meanwhile, the outer voltage loop considers a proportional-integral action. Additionally, the current loop implements an adaptive hysteresis logic in order to operate at a fixed frequency. The closed-loop systems performance is checked via numerical results with respect to step changes in the load, input voltage, and output voltage reference variations.

This article presents a novel simplified method to implement the finite set - model predictive control technique for photovoltaic generation systems connected to the ac network. This method maintains the advantages of the conventional finite set - model predictive control, such as fast response, simple implementation, and easy understanding; but it also eliminates the use of a cost function and hence the weighting factors, instead, it finds the optimal operating state directly from the model and the discrete number of valid states of the converter. Although the proposed algorithm does not compute a cost function, it is able to select the inverter state that minimizes the tracking error by using a hexagonal convergence region. The main advantage of this technique is to reduce the computational cost in 43% of the algorithm that selects the best state, presenting a simple and complete algorithm without compromising the predictive control performance. The proposed algorithm properly operates under various conditions such as changes in the network frequency and changes in the system parameters.

Currently, industry is undergoing an exponential increase in binary-based combinatorial problems. In this regard, metaheuristics have been a common trend in the field in order to design approaches to successfully solve them. Thus, a well-known strategy includes the employment of continuous swarm-based algorithms transformed to perform in binary environments. In this work, we propose a hybrid approach that contains discrete smartly adapted population-based strategies to efficiently tackle binary-based problems. The proposed approach employs a reinforcement learning technique, known as SARSA (State-Action-Reward-State-Action), in order to utilize knowledge based on the run time. In order to test the viability and competitiveness of our proposal, we compare discrete state-of-the-art algorithms smartly assisted by SARSA. Finally, we illustrate interesting results where the proposed hybrid outperforms other approaches, thus, providing a novel option to tackle these types of problems in industry.

One of the main tools for monitoring the condition of high voltage equipment is the measurement of partial discharges (PD). The electromagnetic (EM) radiation originated from this degradation phenomenon can be captured by various types of ultra-high frequency (UHF) antennas carefully designed and optimised for specific frequency bands. However, the presence of environmental noise may limit the use of this technique. Different types of monopole antennas normally used in UHF PD measurement have been evaluated in order to validate the performance of a novel separation technique of EM sources. Accordingly, a new separation technique based on the energy ratio of the captured signals was developed, considering noise interferences. The results revealed that the new technique allows an adequate separation, even when three sources act simultaneously.

El transporte óptimo ha sido tema de interés durante los últimos años [Villani, 2008] por sus aplicaciones en gestión de operaciones y en procesamiento de imágenes. En particular, en [Ekeland and Queyranne, 2015] se plantea un modelo de transporte óptimo para la extracción de un pozo minero a cielo abierto. El objetivo de este trabajo es presentar un modelo similar al descrito anteriormente, pero agregando restricciones de capacidad de extracción. Utilizando el esquema de dualidad de [Pennanen and Perkkiö, 2019] para problemas en el espacio de medidas de Radón, se consiguen condiciones de optimalidad para las soluciones, las cuales generan un pozo. Numéricamente se consiguen resultados para la optimalidad del pozo en el problema discretizado tanto para el problema irrestricto como para el problema con la restricción de capacidad.

Project Management is becoming increasingly crucial in competitive environments such as manufacturing and the service industries. The resource-constrained multi-project scheduling problem (RCMPSP) consists of assigning start times to jobs corresponding to two or more projects that must be executed simultaneously while respecting the precedence between jobs and limited resources. The existing rise in the study of the RCMPSP resulted in numerous works on the topic while proposing different problem features. This thesis presented a study of the RCMPSP and its variants, the base problem is initially described, and the solution methods and benchmarks used to solve RCMPSPs are classified and analyzed. The implications of the RCMPS with practice and its connection are also discussed. In addition, this research analyzed different variants of the problem based on aspects related to jobs, projects, relationships, resources, and time management, and based on the problem variants considered in the reviewed works, a taxonomy allowing (i) the identification and positioning of each RCMPSP variant and (ii) the analysis of the current state-of-the-art of the problem was proposed. Subsequently, it delved into the research of one of the most studied variants in the literature, RCMPSP with local resources (RCMPSP-L), describing the problem in detail and exposing its corresponding mathematical formulation. To solve the RCMPSP-L, a hybrid algorithm that integrates ant colony optimization with a hill-climbing first-improvement algorithm was proposed, and the experimentation process was carried out using a set of instances taken from the MPSPLib. The proposed algorithm presented low variability values, the construction phase provided good-quality solutions, and the local search constantly improved the constructive algorithm, with the results compared with other well-known approaches in the literature, evidencing that for different existing metrics, the proposed approach presents competitive results. Later, new variants of the problem that consider resource mixed accessibility, total resource allocation, delay costs, and resource costs were presented: RCMPSP-MT-W and RCMPSP-MT-W+RC, respectively, both variants described and mathematically formulated. A hybrid algorithm that integrates priority rules, simulated annealing, and tabu search as part of an iterated local search was proposed to solve the variants, where simple perturbation and joint perturbation were proposed as different perturbation intensities as part of the ILS. To validate the model and the algorithm, 36 instances based on the MPSPLib library were modified, and the results show that the application of joint perturbation yields better or equal results than simple perturbation, making greater use of resources with global accessibility, and although it may not be the only factor, the global accessibility of resources may imply greater freedom of allocation, allowing better results to be obtained.

La segmentación automática de lesiones de esclerosis múltiple en imágenes de resonancia magnética es una tarea fundamental para el diagnóstico, el monitoreo de la enfermedad y la evaluación de tratamientos. No obstante, la presencia del efecto del volumen parcial, el solapamiento de intensidades entre tejidos y el alto desbalance de clases, dificultan la segmentación de las lesiones de EM. En esta tesis se propone una nueva función de pérdida, denominada Mahalanobis Distance Loss (MDL), basada en el Mapa de Distancias de Mahalanobis (MDM), también propuesto, que integra información espacial y textural mediante características radiómicas extraídas de la modalidad FLAIR. A diferencia de los mapas de distancia construidos con la distancia euclidiana, el MDM incorpora dependencias estadísticas entre características, capturando mejor las variaciones sutiles en regiones ambiguas cercanas a los bordes de las lesiones. La MDL es combinada con la Generalized Dice Loss mediante un parámetro $\epsilon$ que regula el equilibrio entre solapamiento global y precisión en los bordes. La evaluación en los conjuntos de datos públicos ISBI-MS y MSSEG2016, utilizando una U-Net, demuestra que la función de pérdida propuesta supera a las basadas en mapas de distancia euclidiana, como Boundary Loss y Hausdorff Loss, en métricas de solapamiento (Dice, precisión), de borde (HD95, ASSD) y de detección bajo desbalance de clases (AUC-PR), además de reducir los falsos positivos. Los resultados validan que incorporar información de textura en la función de pérdida mediante el MDM mejora la segmentación automática de lesiones de EM y ofrece un marco prometedor para generalizar estas ideas a otros tipos de lesiones, tejidos y órganos.

El presente estudio aborda el desarrollo de un sistema operativo de respuesta ante emergencias Search and Rescue (SAR) en ambientes marítimos, basado en el uso de Modelos Hidrodinámicos Globales (MHG) y un Modelo Hidrodinámico Local (MHL), específicamente FVCOM. Se diseñaron algoritmos para procesar las salidas de estos modelos y utilizarlas como entrada en el Estimador Múltiple de Deriva de Partículas (EMDP), con el objetivo de delimitar zonas de búsqueda ante eventos de deriva en el mar. La evaluación del sistema se realizó utilizando datos del accidente aéreo del 2 de septiembre de 2011, ocurrido entre Isla Santa Clara e Isla Robinson Crusoe. Los resultados evidencian una buena concordancia entre trayectorias simuladas y observadas, con variaciones asociadas a las condiciones océano-atmosféricas. Se observó mejor desempeño de los modelos influenciados por mareas en condiciones de calma, y de los modelos impulsados por viento cuando este superó los 7 km·h⁻¹. El uso del MHL permitió una resolución espacial fina (hasta 46 m), capturando con mayor precisión la distribución de trayectorias. Este trabajo representa el primer aporte científico en Chile orientado a emergencias SAR marítimas, y propone una herramienta de apoyo eficiente para autoridades responsables en la toma de decisiones.