ABSTRACT. Traditional agricultural pest management relies mainly on chemical pesticides, but their use is linked to ecological damage and the development of pest resistance. Although biological control offers a sustainable alternative, the industrial-scale production of biological control agents is restricted by stochastic biological variables and unpredictable population dynamics. This study aims to address the challenge of predicting the mass production of biological control agents within greenhouses by using artificial intelligence. Using a proprietary dataset, three machine learning architectures (ANN, SVM, and GB) were evaluated to predict the final biological control agents population. For simulation condition 1 specifically, while the ANN and SVM models exhibited negative R² values, the GB model demonstrated superior predictive capacity, achieving an R² value of 0.574 and exhibiting the lowest error magnitudes (RMSE = 592.52, MAPE = 0.142). However, for simulation condition 2, the ANN model performs better with an R² value of 0.704. These results suggest that ANN is a mathematically viable tool for optimizing bio-production processes, enabling real-time population estimation and improving the economic and ecological sustainability of biological control programs.

ABSTRACT. This paper develops a mixed-integer nonlinear optimization model to minimize the operating cost of a distribution feeder, coordinating a hybrid system comprising photovoltaic generation and electric vehicles. The modeling uses energy arbitrage to schedule EV charging and discharging based on electricity prices, photovoltaic generation availability, and network operating constraints over a 24-hour horizon. Results for the IEEE 15-bus radial distribution feeder show that, compared with photovoltaic systems operating without energy storage, integrating EVs allows more effective use of available solar generation and avoids photovoltaic power curtailment. In addition, coordinated EV operation helps alleviate conductor congestion, reduces on-load tap changer operations, and produces a smoother net demand profile.

ABSTRACT. In this work, a mixed-integer nonlinear programming (MINLP) model is compared with its reformulation using a mixed-integer second-order cone programming (MISOCP) relaxation for the optimal siting and sizing of distributed renewable generation and energy storage systems in radial distribution networks. Uncertainty in demand and resource availability is modeled using a stochastic approach based on multiple scenarios and evaluated on a 12-node radial system across two case studies. The results demonstrate that the MISOCP formulation can obtain feasible solutions to the original MINLP model, significantly reducing computational time, improving the objective function value, and maintaining consistency in planning decisions. Consequently, the proposed relaxation constitutes a computationally efficient and technically sound alternative for the planning of distribution networks with high penetration of distributed energy resources.

ABSTRACT. Generator maintenance scheduling (GMS) is a key activity in power system planning. However, the impact of workforce availability constraints and their economic implications has received limited attention. This paper proposes a mixed-integer adaptive robust optimization model with finite adaptability for workforce hiring under uncertainty in labor availability. Maintenance decisions are made here and now, while hiring decisions adapt as uncertainty is progressively revealed. The model is tested on a single-bus system with five thermal units and a twelve-period horizon. Results show that deterministic planning may lead to significant workforce shortages under low-availability scenarios. Although ex-post corrective rules reduce these shortages, they do not guarantee feasibility. In contrast, the adaptive robust model eliminates shortages in all considered trajectories and achieves a lower worst-case cost than the corrected deterministic alternatives.

ABSTRACT. This paper proposes a linear-programming-based under-frequency load shedding scheme to adjust the maximum limit of a continuous UFLS control under severe contingencies. The model employs an aggregated system frequency response as a predictor and incorporates explicit nadir and steady-state frequency constraints. The proposal is validated in DIgSILENT PowerFactory on the IEEE 39-node system. Results show that the method keeps the frequency above operating limits with less load shedding than the reference continuous scheme and prevents the operation of under-frequency generator protections.

ABSTRACT. This paper presents the design and experimental validation of an Internet of Things Cyber Physical System for comparative aeroponic and aquaponic cultivation under controlled environmental conditions. The proposed architecture integrates environmental sensing, embedded monitoring, wireless communication, hydraulic actuation, and real time data acquisition within a Smart Agriculture framework. The system was implemented using Arduino UNO R4 WiFi, ESP32 controllers, industrial communication technologies, and environmental sensors for temperature, humidity, acidity, electrical conductivity, dissolved oxygen, and water level monitoring. A laboratory scale prototype was evaluated during three independent cultivation cycles comprising ninety six observations. Environmental variables were monitored continuously through an Internet of Things acquisition infrastructure. Experimental results showed that the aeroponic subsystem achieved twenty seven percent higher fresh biomass production than the aquaponic subsystem under identical environmental conditions. Statistical validation confirmed significant differences between cultivation systems, while effect size analysis indicated a substantial practical impact favoring aeroponic cultivation. The results demonstrate the effectiveness of the proposed Cyber Physical System architecture for Smart Agriculture applications and establish a scalable foundation for future integration of artificial intelligence, edge computing, and autonomous environmental control

ABSTRACT. Visual object traceability in production environments remains a significant challenge for smart manufacturing. This paper proposes an AIoT architecture for object detection, counting, and remote monitoring using YOLOv11 Nano. The proposed architecture incorporates a confirmed-event counting mechanism that transforms visual detections into persistent historical traceability records. The system was evaluated under varying distances, lighting conditions, multi-background scenarios, simultaneous object presence, and partial occlusions. Experimental results achieved an mAP@50 of 99.5\% on the independent test set, while maintaining correct object identification in representative partial-occlusion scenarios with visually estimated occlusion levels of approximately 30\% to 50\%, demonstrating its potential for visual traceability and intelligent monitoring applications in Industry 4.0 environments.

ABSTRACT. Because unnecessary referrals strain patients and healthcare systems alike, heart murmur screening must rule out healthy cases with confidence, not only flag abnormal ones. This motivates Multi-Zone Attention (MZA), a multimodal, patient- level framework for murmur detection from phonocardiograms. MZA fuses acoustic and clinical features through bilinear fu- sion, integrates up to four auscultation sites via a Cross-Zone Transformer (CZT) with rotary positional encoding, and adapts a pretrained Audio Spectrogram Transformer through Low-Rank Adaptation. On a held-out test set of 442 patients from the CirCor DigiScope Dataset, MZA reaches an F1-score of 0.932 on the murmur-absent class (94.3% precision, 92.1% recall), competi- tive with patient-level baselines from prior work. Post-training calibration sets a conservative decision threshold of 0.385, and temperature scaling reduces the expected calibration error, so negative predictions rely on calibrated probabilities rather than raw confidence. CZT attention correlates strongly with ablation- derived zone importance and follows acoustic content rather than positional identity, indicating clinically relevant, not positional, attention.

ABSTRACT. Evaluating cooperative algorithms in multi-agent systems (MAS) under uncertainty remains an open methodological challenge because most studies consider individual sources of non-determinism in isolation, preventing fair cross-protocol comparison under realistic combined conditions. This paper presents the design and initial validation of a reproducible benchmarking framework for cooperative multi-agent systems operating under combined structural non-determinism. The framework simultaneously models four structural uncertainty sources—switching topologies, link and actuator failures, asynchrony, and partial observability—within a distributed information system domain. Three established cooperative consensus protocols are evaluated across five parameterized scenarios using six protocol-agnostic metrics over 450 controlled simulations (K=30 runs per scenario) with fixed random seeds to guarantee bit-exact reproducibility. The framework consistently differentiates protocol behavior and reveals four empirical observations: (H1) a measurable complexity gradient under nominal conditions, (H2) counter-intuitive convergence acceleration under small-world network partitions, (H3) differential fault tolerance under node failures, and (H4) a performance ceiling for classical consensus protocols under combined structural uncertainty. These findings constitute an initial methodological validation of the framework rather than definitive statistical evidence; formal inferential analysis and comparisons with learning-based approaches are planned for future work. All experimental artifacts will be released as an open Jupyter notebook upon paper acceptance to facilitate independent replication and extension.

ABSTRACT. In aquaculture, capturing consistent optical data is a fundamental prerequisite for monitoring parameters such as chlorophyll-a and turbidity to prevent eutrophication. However, many computer vision systems that are commonly used in outdoor ponds nowadays face challenges related to environmental factors. Such problems include sunlight changes and radiometric noise, and both can considerably affect or reduce prediction accuracy. This work presents the design, electronic implementation, and field evaluation of an autonomous optical sensing system designed for aquaculture environments. The system uses an ESP32 microcontroller connected through I2C to a BH1750 light sensor and an INA219 current monitor. The lighting stage was optimized using a MOSFET to reduce temperature variations. Tests showed a linear relationship (R² = 0.99788) between electrical current and light intensity. Histogram analysis from the Region of Interest (ROI) confirmed the mechanical and optical reliability of the system and registered drops to zero in the blue channel, which in turn confirmed the physical principle of light absorption behavior of chlorophyll-a.

ABSTRACT. Microwave Imaging Technology (MWI) is an easy and practical technology, which is applied to the medical diagnosis, and it represents a non-invasive alternative, that offers medical escort and medical support to prevent, detect, and discover abnormalities or breast tumors, brain tumors or tumors in general, and it helps people, women, men and children, with limited financial resources, due to magnetic resonance imaging (MRI) is expensive and practically unavailable in rural medical units. In this sense, the purpose of this work consists of design, implementation and measurement of a 4 X 4 array antenna operating at 3.5 GHz frequency band as an important part of microwave imaging device, electronic device applied to medical care.

ABSTRACT. This research applies parameters to a DC motor system through a state-space mathematical model, analyzing the system from the initial equations that describe the interaction between the electrical and mechanical domains, highlighting their electromechanical coupling. By applying the Cayley-Hamilton theorem to obtain the state transition matrix, the time response is simplified through the characteristic polynomial, yielding an analytical solution that effectively eliminates cumulative errors. Alongside this, a performance presentation is provided through numerical results and graphs describing the motor's dynamics at startup under a step input, allowing a visual comparison of the transient response and the steady-state performance under different load conditions. The analytical and numerical development, together with the inclusion of Lyapunov stability analysis where a positive definite matrix P ensures a negative derivative strengthens the model by mathematically guaranteeing the system's convergence to its equilibrium point through energy dissipation. This is achieved through user-friendly software implementation and interface functions, which not only validates the viability of the proposed theoretical model and ensures data consistency, but also facilitates the iterative design of more robust control strategies and optimization for applications.

ABSTRACT. Silicon Photomultipliers (SiPMs) have been established as a standard technology in light detection due to their extraordinary sensitivity—enabling single-photon detection—and their null response to magnetic fields. These characteristics make them ideal for critical applications such as Time-of-Flight PET imaging, high-energy physics detectors, LiDAR distance measurement systems, astrophysical applications, and quantum cryptography, among many others. However, their avalanche-based operation (Geiger mode) means that each detection generates heat via the Joule effect. If this self-heating is not properly managed, it leads to adverse consequences such as increased Dark Count Rate (DCR), Optical Cross-Talk (OCT), and gain fluctuations. This work describes an experimental methodology to determine the power dissipation density of a Hamamatsu S13360-3075PE SiPM in a controlled environment. The characterization is performed by exciting the device with an LED, which is driven by a periodic pulsed signal with variable amplitude, frequency, and width. This approach simulates diverse working conditions and provides an empirical quantification of the thermal stress of the device. A power dissipation density of 1.372 mW/cm^2 was determined under the designed experimental setup.

ABSTRACT. Open-circuit fault diagnosis in multilevel inverters is important to improve the reliability of power conversion systems. In cascaded H-bridge inverters, a fault in one power switch can modify the output voltage waveform and cause the loss of voltage levels, affecting inverter performance. This paper proposes a Python-trained neural network implemented on an FPGA for open-circuit fault diagnosis. The proposed method is based on the acquisition and processing of the inverter output voltage, from which the average and RMS values are obtained. These parameters are used as neural network inputs to classify normal operation and fault conditions. Since some faults produce similar effects, the operating conditions are grouped into seven classes. Experimental results show that the proposed system correctly identifies the corresponding fault class.

ABSTRACT. This paper presents the design and implementation of a low-cost DC Line Impedance Stabilization Network (LISN) for conducted electromagnetic interference (EMI) measurements in power converters used in DC microgrids. The calibration results of the proposed DC LISN are presented in terms of insertion loss, isolation, and impedance characteristics. To validate the proposed LISN, a Cuk DC–DC converter for LED lighting applications was selected as the Device Under Test (DUT). Conducted EMI measurements were performed in accordance with the CISPR 15 international standard using an EMSCOPE EMI receiver. The measurement results obtained with the implemented DC LISN were compared with those acquired using an RF-channel oscilloscope, demonstrating the suitability of the proposed LISN for conducted EMI measurements.

ABSTRACT. This study evaluates the technical, economic and environmental sustainability of implementing energy self-sufficiency at the San Antonio de los Buenos WWTP in Baja California, Mexico, through biogas recovery from sewage sludge. Using operational data (95,040 m³/day, 24,710.4 kg VSS/day) and a simplified steady-state Anaerobic Digestion Model (ADM1-based), a Continuous Stirred-Tank Reactor (CSTR) under mesophilic conditions (35°C, HRT = 20 d) was modelled. Results indicate a methane production of 6,774.8 m³ CH₄/day, equivalent to 26,943 kWh/day of electrical energy (40% CHP efficiency) and 30,311 kWh/day of thermal energy (45% efficiency). The energy self-sufficiency index reaches 70.9%, avoiding 82,207 t CO₂e/year. Economic analysis yields a payback period of 6.3 years, NPV of USD 1.0 M and IRR of 14.9%. This study confirms that energy self-sufficiency in medium-to-large WWTPs is technically achievable and financially attractive under current Mexican energy policies.

ABSTRACT. The sustained increase in electricity demand and the need to reduce fossil fuel consumption have driven the development of solutions focused on harnessing renewable energy in residential applications. To this end, an intelligent renewable energy management system was developed based on the integration of electricity generation from photovoltaic and wind sources, along with a battery bank and backup power supplied by the Federal Electricity Commission (CFE). The system integrates a platform for acquiring electrical and temperature data with a Raspberry Pi-based controller capable of real-time monitoring of voltage, current, temperature, and battery bank charge level. It also incorporates an energy management algorithm that prioritizes the use of renewable sources and only connects to the electrical grid when operating conditions require it. Tests conducted demonstrated stable operation throughout the evaluation period. The results obtained support the viability of this architecture for increasing the use of renewable energy and reducing dependence on conventional electricity generation sources.

ABSTRACT. In modern, renewable-based electrical energy generation, power converters are essential for interfacing the system components. In particular, three-phase bridge inverters are needed to interconnect renewable energy resources to the utility grid through a filter. In this sense, the filter design is crucial for reducing the harmonic content of the current waveform injected into the grid, thereby ensuring compliance with industrial interconnection standards. This paper proposes the modeling, design, and parametrization of an LCL (inductive-capacitive-inductive) filter to achieve an effective energy transfer with low distortion, based on frequency analysis using Bode plots. Experimental results expose the filter’s high performance. Additionally, an L and LC filter are analyzed to compare the filtering characteristics of these low-order designs.

ABSTRACT. This paper reports the Control Hardware-in-theLoop (CHIL) validation of a hierarchical control scheme for an ac microgrid formed by two three-phase voltage source converters, local loads, and a grid connection. The emphasis is placed on the implementation itself and on the experimental evidence. The power stage is solved in an RTDS simulator with a 2 µs time step, the control models run at 25 µs on a separate processor card, and the primary control of one converter is executed on an external TMS320F28379D digital signal processor that closes its loop through the digital input and analog output cards of the simulator. The partitioning of the model across processor cards, the port assignment of the GTDI and GTAO cards, and the scaling of the exchanged signals are described in detail, since they determine whether the external loop closes at all. Five experiments cover islanded operation without restoration, voltage restoration, frequency restoration, grid interconnection, and operation with droop slopes obtained from a constrained loss-minimization problem, which reduces the measured total losses by 3.04%.

ABSTRACT. This work presents a control scheme to regulate the input voltage of a system composed of a photovoltaic (PV) module, a boost converter, and a microgrid as a load. The regulation of the converter’s input voltage is due to the fact that the power generated by a photovoltaic system is related to the voltage at its terminals, where a passivity based control (PBC) is proposed as a maximum power point tracker (MPPT) scheme. In the design of the controller, it was considered that only the capacitors’ voltage signals are available, taking advantage of the properties of the system average model, and calculating the inductor current via a state estimator. The closed-loop system exhibits input voltage tracking with all internal signals bounded, so a formal stability proof is presented using Lyapunov theory, validating the above with simulation results.

ABSTRACT. This paper presents the Hardware-In-the-Loop (HIL) validation of a 5-kW bidirectional three-phase active front-end (AFE) converter intended for grid-connected energy conversion applications. A real-time HIL platform based on the Typhoon HIL 402 emulator and a TMS320F28335 digital signal processor (DSP) was developed to evaluate the converter under both rectifier and inverter operating modes. The implemented control strategy regulates the DC-link voltage and grid currents while enabling bidirectional power flow.

ABSTRACT. Electric railway systems present challenges related to power quality and the integration of alternative energy sources. This has motivated the development of power-quality conditioners, renewable-energy interfaces, and advanced control techniques. For this reason, this paper presents the modal modeling and control of a three-phase active rectifier connected to a 2×25 kV railway power supply system. The main objective is to address the fundamental question of how to connect this converter to such a railway system. A modal model is used, showing that the converter dynamics can be represented in terms of two decoupled current modes. The first mode is associated with power exchange between the converter and the railway system, whereas the second redistributes current among the converter branches. Based on this model, current controllers are designed for both modes, and a DC-link voltage controller is implemented. Simulation results show effective regulation of the DC-link voltage while independently redistributing currents among the branches. The results obtained provide a control framework for operating these converters, which could be used for power-quality conditioning or as an interface for alternative energy sources in 2×25 kV railway power supply systems.

ABSTRACT. 2×25 kV electric railway systems, such as the Tren Maya in Mexico, employ a center-tapped secondary supply to feed the catenary and the negative feeder. Due to the spatial distribution of railway loads and the characteristics of the supply network, unequal loading conditions may arise between the two sections of the secondary winding, as well as in reactive power demand. These conditions result in a non-uniform distribution of currents and power seen by the traction substation. This paper proposes using a three-leg inverter connected to a 2×25 kV railway system to compensate for reactive power and balance active power between the two supply branches. The proposed method exploits the additional degree of freedom afforded by the third converter leg to redistribute power between the two sections of the secondary winding without altering the net power transferred to the railway system. Simulation results show that the proposed strategy significantly reduces the active power imbalance between the supply branches while compensating for the reactive power demanded by the system. The results show that the converter can serve as an active interface between the railway network and the traction substation, reducing the power imbalance among the secondary winding sections and compensating for the reactive power seen by the substation.

ABSTRACT. Partial Power Processing (PPP) has proven to be an effective strategy for increasing the efficiency of DC/DC converters by reducing the fraction of power processed by the conversion stage. As an extension of this concept, Reduced Redundant Power Processing (R2P²) enables the development of new architectures with lower losses, high transformation ratios, and greater design flexibility. This paper presents a systematic methodology for deriving a family of buck-boost converters based on the R2P² concept. The methodology establishes synthesis rules that allow for the generation of different topologies by combining basic conversion structures, while preserving the advantages of partial power processing. Furthermore, voltage conversion relationships are derived, and a theoretical analysis of the proposed configurations is performed. The results demonstrate that the methodology constitutes a general tool for the development of high-efficiency Buck–Boost converters geared toward energy storage applications, renewable energy, and direct current systems.

ABSTRACT. Thermal analysis is a critical aspect of power electronic converter design, as proper thermal management directly affects system reliability, efficiency, and power density. Since photovoltaic microinverters are typically installed in outdoor environments, they are exposed to varying and often harsh operating conditions, making effective thermal design even more essential. As the first stage of a microinverter is commonly designed using a full-bridge inverter, this paper presents the analysis and thermal behavior of the power semiconductors using three different heat sink geometries implemented on the printed circuit board. Simulations and experimental validation are performed to evaluate semiconductor thermal behavior under these conditions. The results show that the power converter can be implemented without an external heat sink when an adequate geometry with thermal vias is designed on the printed circuit board.

ABSTRACT. This paper presents the modeling and validation of short-circuit currents in a 750 V DC traction substation with a 12-pulse rectifier. Since high load currents in DC railway systems make remote faults difficult to distinguish from nominal operation, accurate current estimation is relevant for protection studies. First, a method based on the current factor specified in UNE-EN 50328 is used to estimate the steady-state short-circuit current for the most severe terminal fault and is validated against MATLAB/Simulink, with an error of 8.96%. Then, a second-order analytical model is used to estimate the transient and steady-state current response for faults at different locations along the traction line. The model is compared with MATLAB/Simulink simulations in terms of steady-state current, peak current, time to peak, and rate of change of current. For the terminal fault, the errors are 0.065% for the steady-state current and 1.935% for the peak current. Across all fault locations, the maximum errors are 7.715% and 7.779%, respectively. Results show that both methods are suitable depending on the required level of analysis, accuracy, and detail.

ABSTRACT. The integration of distributed generation systems introduces high-frequency switching noise and dynamic electrical variations, complicating the accurate classification of Power Quality Disturbances (PQD). Traditional feature extraction methods often face a trade-off between temporal-spectral resolution and computational cost, hindering their deployment in Edge Computing architectures. This paper proposes a lightweight machine learning methodology based on a Multilayer Perceptron (MLP) for classifying seven PQD categories, utilizing empirical signals. To prevent class overlapping caused by temporal dilution in algorithms, low frequency disturbance events were standardized to 10-cycle windows. Feature extraction was performed using the Modified Stockwell Transform (MST) to preserve time-frequency morphology, followed by Principal Component Analysis (PCA) to drastically reduce the dimensionality of the feature vector while retaining 95% of the variance. The proposed MLP classifier, trained on this compressed PCA space, achieved a global empirical accuracy of 98%. The results demonstrate that standardizing evaluation windows combined with MST and PCA addresses physical ambiguities between classes such as Sags and Interruptions, presenting a highly efficient algorithmic pipeline viable for real-time inference on resource-constrained microcontrollers.

ABSTRACT. This work presents a methodology for real-time dynamic reconstruction of a direct current machine based on a reduced electromechanical model and experimentally identified parameters. Unlike conventional approaches that rely on assumed machine parameters or offline identification procedures, the proposed methodology directly determines the electrical and mechanical parameters required by the model, including armature resistance, field and armature inductances, electromechanical coupling coefficient, rotational inertia, and equivalent mechanical damping. A real-time acquisition platform developed in LabVIEW continuously measures the electrical and mechanical variables required to numerically estimate field current, armature current, electromagnetic torque, and rotor speed during operation. The proposed methodology is validated under no-load, partial-load, and full-load conditions by directly comparing measured and estimated variables. Results demonstrate close agreement between measured and reconstructed responses, confirming the capability of the identified model to represent the electromechanical behavior of the machine under different operating conditions

ABSTRACT. This paper evaluates the use of Lyapunov exponents to assess transient stability. They have recently been proposed to assess stability using the data from a time-domain simulation program, or from phasor measurement unit devices, in off-line or real-time modes respectively. This document presents the basics of Lyapunov exponents and, due to its simplicity and flexibility, studies the model-free approach, which is applicable to both, real-time measurements and simulation results from a time-domain program. Results in the 3-machine 9-bus IEEE test system shown that the Lyapunov Exponents criterion requires many additional improvements to be suitable for transient stability assessment.

ABSTRACT. This contribution addresses the time-domain assessment of the harmonic state of a power system containing nonlinear components and nodal current injections to represent the harmonic contamination during system operation. The harmonic state estimation is evaluated using the unscented Kalman filter algorithm based on the unscented transform. This allows the consideration of noise in the measurements of physical variables taken from the system, which can be voltage and current waveforms and signals used to evaluate the state estimation. Case studies are applied to transient conditions in the generation of harmonics, and varying harmonics are also analyzed. The total harmonic distortion is evaluated by applying the discrete Fourier transform to convert time-domain data to frequency-domain data to obtain the magnitude and phase angle of the harmonic content. These results can be compared against the power quality standards, and the resulting magnitudes and phase angle can be applied to design and control harmonic filters to mitigate the voltage and current waveforms harmonic distortion. The proposed unscented Kalman filter harmonic state estimation method is successfully validated against the timedomain nonlinear power system response.

ABSTRACT. The Internet of Things (IoT) has enabled the seamless connection of multiple devices, allowing for remote monitoring and control of various environmental variables. In the context of water consumption, IoT technologies have been successfully applied to measure water usage in real-time, enabling leak detection and reducing water waste. Furthermore, analyzing household water consumption data provides valuable insights into consumption patterns, aiding in the development of strategies for the efficient and sustainable management of water resources. This paper presents an IoT-based system for monitoring and controlling water consumption. The system collects water usage data via an IoT device and stores it in a \texttt{MySQL}~database. A Front-End interface is developed to visualize the water consumption data for users, while the Back-End system handles data storage and reporting, facilitating the identification of normal and abnormal consumption patterns. An analysis of water usage for a single household is performed using Python. The proposed system leverages several open-source platforms and utilizes the \texttt{ESP8266}~microcontroller to create a low-cost IoT architecture.

ABSTRACT. This paper presents a study to identify key frequency bands, which allow the classification of distinct microearthquakes that encompass: long-period (LP) events, volcano-tectonic (VT) earthquakes, tectonic (TC) earthquakes, and volcanic tremors (TR). The proposed approach employs a feature extraction methodology that capitalizes on the energy associated with the approximation (cA) and detail coefficients (cD) derived from multilevel 1-D wavelet decomposition. Event classification is performed using the Decision Tree (DT) fortified with the feature selection stage that allows the algorithm’s efficiency to be improved. In order to evaluate the efficacy of this characterization methodology, a dataset from the Llaima volcano was used. The findings of this paper underscore that the characterization of microearthquakes can be notably defined by specific wavelet detail coefficients, which correspond to distinct frequency bands, which are the range between the lower (FL) and upper (FH) frequencies of a given interval, that is, fb in (FL, FH) Hz. Specifically, cD3, cD5, and cD7 include fb in (6.25, 12.5) Hz, fb in (1.56, 3.12) Hz, and fb in (0.39, 0.78) Hz, respectively.

ABSTRACT. Automated testing in Continuous Integration / Continuous Delivery (CI/CD) coordinates web workflows, Application Programming Interface (API) contracts, visual baselines, and telemetry across heterogeneous tools, yet empirical research rarely isolates the test automation architecture as the independent variable. This paper evaluates a Test-Oriented Microkernel Architecture (TOM) against GTAA_BASELINE, a layered architecture inspired by the International Software Testing Qualifications Board (ISTQB) Generic Test Automation Architecture (gTAA), under a frozen 50-run CI protocol in which both arms share scenarios, assertions, contracts, test data, and failure buckets. TOM achieved 15,073/15,083 functional passes (99.93%) versus 15,303/15,350 (99.69%) for GTAA_BASELINE, with failures in 1/50 runs versus 31/50. Failure-bucket telemetry recorded 7 TOM events versus 109, and byte-identical visual contracts produced 0/4,509 TOM failures versus 15/4,089. API and Gatling reliability were equivalent (1,650/1,650 API passes and 0/150 request failures per arm). Maintainability favored TOM: 73.60 versus 141.22 average lines of code (LOC) and 8.18% versus 16.77% duplicated code. TOM reduced recurrent failure exposure and maintenance burden without degrading API or performance-test reliability; latency, mobile portability, modifiability, and extensibility remain unresolved.

ABSTRACT. Ensuring the integrity of the evaluation environments is critical for reliable online assessments. Yet, programming-contest and exam labs that still require one bootable USB drive per workstation do not scale. Every last-minute policy change means re-flashing all of them by hand. Huron OS, the operating system purpose-built for exam-integrity enforcement, appeared architecturally tied to USB media until its own official build tooling revealed the real cause: the kernel already ships every network driver it needs, disabled by nothing more than a single build-time switch. This paper flips that switch and adds a small set of additive patches --- HTTP-based system retrieval, on-demand exam-software fetching, and cross-reboot persistence --- letting an unmodified Huron OS boot entirely over the network via iPXE, an open-source boot firmware built on DHCP, TFTP, and HTTP, while enforcing its native exam directives (site allowlist, USB lockout, scheduled modes) exactly as before. The pipeline boots cleanly to a full desktop in a QEMU-based simulated lab, and again on a real-hardware pilot built from a Raspberry Pi master, an isolated MikroTik segment, and a physical client laptop, confirming that PXE boot, directive enforcement, and on-demand software all survive the jump from simulation to metal.

ABSTRACT. Vulnerability management represents a critical challenge for organizations, since the number of detected vulnerabilities often exceeds the available remediation resources. Traditional approaches rely mainly on the Common Vulnerability Scoring System (CVSS), which measures the technical severity of a vulnerability but does not necessarily reflect its estimated likelihood of exploitation. This limitation may lead to inefficient prioritization of remediation efforts. This article presents a device-level vulnerability prioritization model that integrates CVSS severity, exploitation probability estimated through the Exploit Prediction Scoring System (EPSS), intelligence from the Known Exploited Vulnerabilities (KEV) catalog, and the operational context of assets. Internal variable weights are estimated through a hybrid scheme based on Principal Component Analysis (PCA) and entropy, while expert-defined boosts reinforce KEV membership, exploit availability, and high EPSS signals. Priority levels are then assigned using adaptive percentile thresholds. The proposed model is implemented through an automated pipeline and evaluated on data from a real organization comprising 197 devices and 28,886 device--CVE pairs. CVE refers to the Common Vulnerabilities and Exposures identification system, while each device--CVE pair represents the presence of a CVE-identified vulnerability on a specific device. The results show a weak Pearson correlation of 0.063 between CVSS and EPSS, indicating that technical severity alone is insufficient to represent exploitation-related signals. In addition, 98.5\% of vulnerabilities with CVSS scores of 9.0 or higher present EPSS scores below 0.10. The comparison with a ranking based solely on CVSS reveals considerable divergence, with a Spearman correlation of 0.43: 20 of the 30 devices identified as high-priority by the proposed model fall outside the top 30 positions in a severity-based ranking. Overall, the proposed approach offers a practical, reproducible, and decision-support-oriented mechanism to strengthen vulnerability risk management.

ABSTRACT. Industrial Internet of Things (IIoT) technologies have significantly improved remote monitoring and data integration in industrial environments. However, many testing laboratories continue to rely on legacy equipment that lacks native connectivity, limiting data availability and centralized supervision.

This paper presents the design and experimental validation of an IIoT architecture for the remote monitoring of an automotive testing laboratory, enabling the integration of legacy equipment without requiring modifications to its original control systems. The proposed architecture organizes data generation, acquisition, communication, integration, management, and visualization into independent functional layers to promote modularity, scalability, and interoperability.

A prototype implementation was deployed in a real laboratory using industrial testing equipment to evaluate the end-to-end operation of the proposed architecture.

Experimental results validate the proposed layered architecture as an effective framework for integrating legacy equipment into IIoT-based monitoring systems, enabling the initial modernization of automotive testing laboratories through the progressive adoption of Industry 4.0 technologies.

ABSTRACT. This paper presents the design and experimental evaluation of a high-efficiency LED lighting system for the hydroponic cultivation of lettuce (Lactuca sativa) in a Deep Water Culture (DWC) system. The proposed system employs SunPlus 2835 series LEDs from Lumileds and incorporates far-red radiation into the spectral configuration. Light output is regulated using an ESP32 microcontroller through pulse-width modulation (PWM) at 10 kHz, enabling precise control of the electrical power supplied to the LEDs. An eight-week comparative experiment was conducted using a commercially available lighting system as a reference. In both treatments, the photosynthetic photon flux density (PPFD) was maintained at 250 μmol·m⁻²·s⁻¹ by adjusting the electrical power and lamp height. The pH and electrical conductivity (EC) of the nutrient solution were monitored throughout the cultivation period. System performance was assessed based on dry biomass, relative chlorophyll content measured using the SPAD index, and photosynthetic photon efficiency (PPE), expressed in μmol·J⁻¹. The experimental results indicate that the proposed lighting configuration enhanced biomass production and improved the biomass-to-power ratio compared with the reference system. These results support the potential of spectrally optimized LED systems as scalable lighting solutions for hydroponic cultivation and controlled-environment agriculture.

ABSTRACT. This paper presents a comprehensive comparative analysis between a Takagi-Sugeno (T-S) Parallel Distributed Compensation (PDC) controller and the Super-Twisting Algorithm (STA) applied to a 3-Degree-of-Freedom (3-DOF) anthropomorphic robotic manipulator. The control of such systems is challenged by highly coupled nonlinear dynamics, centrifugal forces, and Coriolis effects. The study evaluates the intrinsic trade-off between kinematic tracking precision, measured via the Mean Squared Error (MSE), and the energetic control effort, quantified by the Integral of the Squared Value (ISV). Furthermore, to validate the real-world viability of these strategies on embedded architectures such as the ESP32 microcontroller, the algorithms are evaluated under discrete-time execution constraints (Ts = 5 ms). This includes the integration of a combined Luenberger and Sliding Mode Observer (SMO) to mitigate the severe quantization noise introduced by physical incremental encoders. Results demonstrate that while the STA achieves superior tracking accuracy, the explicitly discretized PDC approach optimizes energetic demands, highlighting the necessity of feedforward compensation for trajectory tracking.

ABSTRACT. This work presentes two single-antenna radio-frequency radars are for the detection of unmanned aerial vehicles equipped with analog First Person View (FPV) video transmitters in the 5.8 GHz band. In the first prototype, the received power is measured with a Software Defined Radio (SDR) and is filtered by an adaptive one-dimensional Kalman filter, while false alarms produced by wideband OFDM emissions, such as Wi-Fi and digital video links, are rejected by a spectral prominence test and an asymmetric temporal filter implemented as a finite state machine. In the second prototype, a low-cost analog video-receiver module driven by two microcontrollers is used, so that the 80 analog FPV channels are swept through a fractional-N phase-locked loop while the video synchronization is verified. In both cases the detection of analog FPV transmitters and the immunity against OFDM interference are demonstrated experimentally.

ABSTRACT. The Minimum-Essential-Parts (MEP) algorithm formalizes sequential control through internal memories, transition conditions, and Boolean output functions. This study evaluates its transferability to electromechanical automation by comparing two implementations of the same three-actuator, 16phase sequence. One prototype was developed on an Alhambra II FPGA using Icestudio, while the second employed an Arduino Mega programmed in Ladder logic. Both systems used position sensors, H-bridge drivers, and DC motor-based actuator emulators to reproduce the extension and retraction states of pneumatic cylinders without compressed air. The comparison examines Boolean-function derivation, memory allocation, output activation, debugging effort, assembly feasibility, and mechanical limitations. Results confirm that the MEP structure remains unchanged across both platforms because it depends on discrete system states rather than actuator technology. The FPGA implementation offers closer correspondence with digital hardware, whereas the Arduino-Ladder solution provides greater accessibility for PLC-oriented automation. Thus, MEP enables systematic, platform-independent design of complex industrial sequential systems.

ABSTRACT. The implementation of Distributed Generation (DG) technologies is key to energy decarbonization, but its infrastructure relies heavily on material-intensive supply chains. Industrial Symbiosis (IS) provides a pathway to mitigate these life cycle burdens by diverting secondary materials and industrial by-products for use as substitutes in power plant components. However, practical tools for operationalizing and quantifying these exchanges are still scarce. This paper proposes a standardized quantitative framework for relating waste flows from DG plants on a kilogram-to-ton scale by specifically tracking material flows. The methodology assesses displaced environmental impacts by comparing quantified waste diversion volumes with the life cycle footprints of traditional primary materials. The model was tested using plant symbiosis scenarios that included critical structural components such as metal alloys. Preliminary results indicate that a localized matching strategy can successfully divert a significant amount of industrial scrap, resulting in a direct, quantifiable reduction in global warming potential per kW installed by eliminating the need to manufacture virgin raw materials. By moving from theoretical circularity to an operational, scale-specific method, this framework provides energy managers and regional planners with a tangible tool to assess the environmental and material viability of industrial symbiosis before implementation.

ABSTRACT. The growing integration of distributed energy resources (DERs) into active distribution networks provides controllable injections that can be operated to manage line loadability and relieve congestion. This paper formulates line-loadability management as a network need and evaluates DERs as the flexibility resource that meets it, using Proximal Policy Optimization (PPO) in two roles: a sizing agent that determines the rated active and reactive power of a set of DERs whose locations are fixed by a loadability-based ranking, and a dispatch agent that operates the sized DERs according to the observed line loading. Loadability is quantified through the Index of Utilization of Lines (IUL) and the Overload Lines Index (OLI). On an integrated transmission and subtransmission/distribution test system (total SDL load: P=251.9 MW, Q=58.2 MVAr at FD=1.0), the learned dispatch policy reduces the IUL by up to 17.64% under the most demanding profile while eliminating the pre-existing line overloads and restoring a fully feasible operating point. A Bayesian hyperparameter search using Optuna is applied to the sizing agent; the control agent hyperparameters are selected from a systematic comparison of five PPO configurations, with results reported as the best-performing policy among all evaluated training runs.

ABSTRACT. This paper presents the modeling and control of a wind energy generation system based on a permanent magnet synchronous machine, interconnected through two converters coupled by a direct current (DC) bus: one on the machine side and one on the grid side. The DC bus is connected to a DC microgrid; on the AC side, the grid-side converter connects to the 220 V electrical grid. The electrical and mechanical model of the machine, the DC bus model, and the grid-side current loops are developed, including the decoupling between the direct and quadrature axes in both converters. The current and speed loops are tuned following Nyquist stability levels, setting for each loop the crossover frequency, the decibels, the phase margin, and the desired overshoot. The system is subjected to a step-load connection test of three resistive loads, evaluating the power demanded by the load and the power supplied by the machine and by the grid at each stage. The results show how the grid compensates for the power deficit when the machine, operating at reduced speed and torque, cannot cover the connected load.

ABSTRACT. —In this work, an open-source tool for wind energy project sizing is presented to support renewable energy implementation. Wind energy is a clean, renewable, and costeffective energy source that reduces greenhouse gas emissions, it requires accurate resource assessment tools for its planning and implementation. The developed application implements the Energy Pattern Factor Method (EPFM), Empirical Maximum Likelihood (EML), and Rayleigh distribution to estimate wind resource characteristics and calculate key technical and economic indicators, including the Annual Energy Production (AEP), Capacity Factor (CF), and Levelized Cost of Energy (LCOE). The tool was applied to compare three locations through an interactive graphical interface. Among the evaluated sites, Juchitan de ´ Zaragoza, Oaxaca, Mexico, exhibited the highest wind energy potential, with an estimated AEP of 28.9 GWh/year and an LCOE of 25.8 [$/MWh], assuming a 25 year project lifetime and a 5.5 MW Nordex N169/5.X wind turbine using manufacturer specifications [1]. The proposed tool provides an accessible framework for preliminary wind energy project assessment and supports informed decision-making for renewable energy deployment.

ABSTRACT. Abstract—This paper implements the Prony method on a digital signal processor (DSP) to identify high-frequency oscillatory modes in electric power systems with power electronic converters, evaluated under a hardware-in-the-loop (HIL) scheme. The system under study, composed of a voltage source converter and a transmission line terminated in a short-circuit impedance, is modeled in the Opal-RT OP4510 real-time simulator, while the Prony algorithm is executed on a TMS320F28379D DSP. An offline parametric study evaluates the influence of sampling frequency, number of samples, and number of estimated modes on detecting an unstable mode analytically set at 2500 Hz. Best performance is obtained when the observation window covers one full cycle of the fundamental frequency. The algorithm is then implemented in real time, and three experimental tests confirm detection of the unstable mode, with execution time dominated by the polynomial root-finding routine. The results confirm the feasibility of Prony-based monitoring for stability applications, where the estimated frequencies and damping factors can trigger protection or operating-mode changes that maintain secure operation.

ABSTRACT. A passivity-based control strategy for a non-inverting step-up/step-down DC--DC converter is proposed and thoroughly evaluated. The control architecture consists of a hierarchical scheme comprising an inner current-control loop and an outer voltage-regulation loop. The inner loop is designed using passivity-based control principles, incorporating energy shaping and damping injection to ensure stable tracking of the involved states. To enhance robustness to parameter variations and modeling inaccuracies, an immersion and invariance-based uncertainty estimator is integrated into the control scheme, providing effective compensation for unknown disturbances and uncertainties. The outer loop is designed as a proportional-integral controller to achieve precise output-voltage regulation and satisfactory steady-state performance. The proposed control strategy is validated through numerical simulations, demonstrating accurate reference tracking, fast dynamic response, and strong robustness in the presence of parameter uncertainties. The results confirm the effectiveness and reliability of the proposed approach for high-performance DC--DC power conversion applications.

ABSTRACT. This paper presents a low-frequency modulation strategy combined with an output voltage regulator for a single-phase five-level NPC inverter. The proposed modulation generates a staircase output voltage waveform with twelve voltage transitions per fundamental period. The modulation strategy also enables natural DC-link voltage balancing while eliminating selected low-order harmonics. The switching instants are defined by three angles per quarter cycle, which are obtained through a nonlinear optimization algorithm to eliminate the third and fifth harmonics while achieving the desired fundamental voltage magnitude. Polynomial approximations are used to select the switching angles for different fundamental amplitudes through a PI control. The proposed system regulates the inverter rms voltage under load and DC-link voltage variations while maintaining capacitor voltage balance. Simulation results validate the performance of the proposed approach where an LCLC filter attenuates the remaining harmonics at the load voltage.

ABSTRACT. This paper presents a passivity-based control (PBC) scheme with an integrated barrier function for the output-voltage regulation of a quadratic boost converter (QBC), which provides a high voltage gain from a low-voltage source. Instead of the interconnection-and-damping-assignment PBC, a classical indirect PBC of the input-inductor current is adopted, which circumvents the non-minimum-phase nature of the output voltage. The main contribution is the integration of a logarithmic barrier function into the current loop so that, within a single energy-based framework, the scheme guarantees both stability through passivity and a hard limit on the inductor current, and provides a soft start. It is shown that a standard (non-recentered) barrier avoids the windup-induced cancellation that defeats the recentered form and that, because the implemented barrier term can only deliver a bounded corrective action—its denominator is floored to avoid the singularity—the current reference must additionally be bounded through an anti-windup mechanism in the outer voltage loop; an explicit design window for this bound is derived. Switching-level simulations of the converter in MATLAB/Simulink (Simscape) for a 12 V to 100 V conversion confirm exact voltage regulation and a current that never exceeds its 8 A safety limit—reaching at most 7.95 A during start-up and a sustained overload, while the classical PBC reaches 10 A—together with a smaller voltage overshoot upon load removal.

ABSTRACT. This work presents a study and analysis of load sharing among DC-DC power electronic converters that are part of a Direct Current (DC) microgrid, which are connected in parallel to meet increasing load demand. The proposed strategy is based on the implementation of a modified PID controller in the voltage control loop, to which a virtual impedance loop is added to improve its performance when converters are connected in parallel. This control strategy enhances the stability and disturbance rejection capability of the microgrid, while also compensating for the effects caused by differences in electrical parameters and variations in operating conditions, thereby achieving a more uniform distribution of current and power among generation sources.

ABSTRACT. This paper presents a comparative analysis of two single-phase multilevel inverters capable of generating seven output voltage levels from a three-capacitor divided DC link. The considered topologies are the asymmetric T-type inverter and the two-leg neutral-point-clamped NPC inverter which are evaluated under identical operating conditions, including the same DC-link voltage, rated power, level-shifted sinusoidal pulse-width modulation, switching frequency, semiconductor devices, and load. The comparison considers the number of required components, conduction and switching losses, output-voltage total harmonic distortion (THD), semiconductor voltage stress, and conversion efficiency, under balanced DC-link capacitors. Power losses are estimated through simulation using semiconductor models derived from the manufacturer datasheets. The results identify the main advantages and limitations of each topology, providing useful information for practical design of these power converters.