Bearing faults are a leading cause of failures in industrial rotating machinery, making their early and reliable detection essential to reduce downtime and maintenance costs. A Hilbert-Huang Transform (HHT) based methodology is proposed for early bearing fault detection. The Empirical Mode Decomposition (EMD) isolates fault-related Intrinsic Mode Functions (IMFs), which are selected automatically using the Kurtosis-RMS (KR) criterion. Envelope Spectrum analysis is then applied to the reconstructed signal for diagnosis. The proposed method achieved 94% accurate IMF selection and over 85% fault identification accuracy across 18 signals from the CWRU, TianYau Wu, and MFPT databases. Compared with conventional FFT and Envelope Spectrum techniques, the approach improves early fault detection under noise and variable load conditions while reducing manual analysis time.
Building smart cities requires well-defined public policies. The use of new technologies allows improving the quality of life of their inhabitants, but the uncontrolled growth of large metropolises slows down sustainable urban development. It is necessary to identify clusters that prioritize the attention of structural problems. The present work refers to the United Nations (UN) Sustainable Development Goals (SDGs) 6 and 11, related to the consumption of drinking water and housing, based on spatial data that make it possible to identify critical patterns to guide public policies. The objective is to analyze the spatial autocorrelation of water use and housing in 2469 municipalities of the Mexican Republic, identifying relevant clusters for sustainable development. The database of the 2020 Population and Housing Survey of the National Institute of Statistics, Geography and Informatics (INEGI) was used, transforming the variables TVIVHAB, OCUPVIVPAR, VPH_AGUADV, among others, to correct biases. Exploratory Spatial Data Analysis (ESDA) is carried out, which includes histograms, thematic maps, and scatter plots. The Moran's index revealed a positive autocorrelation of 0.78 for population density and high clusters surrounded by high neighbors in 28 percent of the municipalities identified through Local Indicators of Spatial Association (LISA). Thematic maps highlight critical areas of low service coverage, underlining the need for integrated policies to improve water and housing infrastructure, which contribute to sustainable urban development in accordance with what a Sustainable Urban Development policy implies from the perspective of the smart cities paradigm
This work presents the design, implementation, and experimental validation of a fuzzy logic controller (FLC) for a two-wheeled balancing robot (TWBR). The controller is structured as a PD scheme, where the proportional action is determined through fuzzy inference using seven linguistic input values, while the derivative action is computed conventionally. The output of the fuzzy system is obtained through a weighted average defuzzification method and converted into PWM signals to drive the DC motors via an H-bridge circuit. The proposed methodology was implemented on an ESP32-WROOM-32 embedded platform, integrating real-time data acquisition from an IMU (accelerometer and gyroscope) to estimate the tilt angle of the robot. Experimental tests were conducted under two scenarios: one without external disturbances and another with manually applied perturbations. Results demonstrate that the fuzzy controller maintains stability within +/- 2 degrees of the setpoint in disturbance-free conditions and successfully recovers equilibrium within a range of 75 degrees to 105 degrees under perturbations. These findings validate the robustness, adaptability, and efficiency of the proposed fuzzy control approach, making it a suitable alternative for embedded robotic applications where stability and real-time response are critical.
This paper presents a software platform developed in LabVIEW, based on a state machine model, for controlling the positioning and angular trajectory of a PVTOL (Planar Vertical Take-Off and Landing) system. Communication with the experimental platform was implemented using Arduino UNO R4 microcontrollers, integrating low-cost distance sensors and a 3DM-GX1 IMU. The data was acquired at 60 Hz via I2C and USB protocols. The software includes signal logging and controller response features for subsequent analysis. Experimental validation, using a classic PID control, demonstrated stable performance of the PVTOL, with no significant oscillations and robustness against external disturbances.
Smart agriculture can contribute to ensuring food security and a sustainable society. However, many rural areas lack access to Internet of Things (IoT) sensor networks and low-cost renewable energy sources, which can decrease their food productivity. To address these challenges, triboelectric-electromagnetic hybrid nanogenerators offer a technological solution for harvesting wind, water flow, and raindrop energy, converting it into electrical energy to power IoT sensors or serve as self-powered devices. Herein, we report recent advances in the development of triboelectric-electromagnetic hybrid nanogenerators for smart agriculture sensing. In addition, the performance and applications of these hybrid nanogenerators in smart agriculture are discussed. Finally, perspectives on the design, materials, and reliability of these nanogenerators are presented. The use of artificial intelligence and digital twins can help to improve the performance and reliability of hybrid nanogenerators for applications in agriculture 4.0.
In a century in which economic, scientific-technical, cultural, political and environmental circumstances demand greater and better energy consumption, new systems are necessary where various technologies and strategies are addressed to maximize energy efficiency and reduce the environmental impact. The proposal in this work is an energy storage system for houses and buildings with an automated control system in which artificial intelligence intervenes with the ability to control the loads connected to the electrical network of the house or building, by which the energy obtained by the system is stored in batteries so that this energy can be used at any time, with its respective transformation process from direct current to alternating current.
In Mexico, the widespread use of liquefied petroleum gas (LPG) in rural areas poses significant safety risks, accounting for 35.86% of domestic incidents reported by CENAPRED (20102020) [1]. This paper introduces “SafeGas,” an Internet of Things (IoT) prototype designed for early LPG leak detection, integrating gas (MQ-6), pressure, and temperature sensors (MAX6675 with Type K thermocouple) alongside secure wireless communication via LoRa (Long Range) with AES-256 encryption and 3G (third-generation mobile networks). The prototype aims to monitor conditions in real time and trigger alerts through a dual-verification system (pressure + gas concentration), addressing key limitations of conventional leak detection methods. The development follows the IoT Development Lifecycle (IDL), structured into six phases: (1)Requirements: Defining objectives. (2)Design: System architecture. (3)Implementation: Hardware, firmware, and encryption. (4)Testing: LoRa range validation, false-positive rate assessment. (5)Deployment: Field installation in rural settings. (6)Maintenance: Firmware updates and performance optimization [24]
This paper presents a comparison of technologies in Medium Voltage Panels that exist in the electricity market. The manufacturing industry that has had facilities for more than 3 decades requires modernizing its electrical systems and energy distribution with efficient medium voltage panels. Today, there are gas-insulated technologies that allow agile protection and selectivity in the coordination of protections. The use of gas insulation is widely used for industrial installations that require optimizing electrical safety and human safety. Based on the above, the comparison of two insulation media is presented, considering the durability of materials, cost, size and long-term maintenance. The purpose is to contribute to the design of safe facilities, considering that the manufacture of this equipment is already a reality in the electricity market. It also tries to contribute to the technical knowledge of engineers who make decisions in projects of this nature, such as the engineers in charge of the Auxiliary Services of electrical maintenance and the Supervision of internal Operation of the Industry in selecting one of the two technologies for a better use in the long term or during its electrical operation and that can be visible in the electrical model for electrical studies that can allow an agile decision-making of the technical personnel in a possible integration of primary electrical equipment.
This paper presents a fuzzy inference system designed to optimize purchase decisions by integrating three key variables: Income, Consumption, and Sales. The proposed model uses triangular membership functions with narrow bases to approximate singleton behavior, enhancing the precision of fuzzy set representation. The inference process is based on a rule base derived from historical data analysis and expert knowledge from materials planning personnel. Defuzzification is performed using the weighted average method, yielding crisp percentage values for purchase orders ranging from 0% to 100%. The system's performance was evaluated through surface plot analysis, which revealed nonlinear and interdependent relationships between the input variables and the purchase output. Results demonstrate that the approach provides accurate and adaptive recommendations for inventory control, offering a valuable decision-support tool for optimizing resource allocation and maintaining stock levels in dynamic operational environments.
The Entropic Associative Memory (EAM) is a cognitively inspired computational model that combines probabilistic recognition with an efficient pattern storage and retrieval architecture. Previous evaluations, limited to small domains of about ten classes and a few thousand instances per class, showed promising results in classification and reconstruction. This work extends those studies in two ways. First, we assess scalability using Googles Quick, Draw! dataset, which includes over 300 classes and 100,000 instances per class. Second, we test EAMs rejection capability by storing only half of the classes and evaluating performance on the full dataset. Results confirm EAMs adaptability to large-scale scenarios and its effective rejection of novel stimuli, underscoring its potential as a robust and explainable AI model.
This work presents the application of the White Shark Optimizer (WSO), a bio-inspired metaheuristic, for the automatic tuning of kinematic controllers in differential-drive robots. Unlike traditional approaches that rely on manual tuning or deterministic optimization, WSO leverages global search strategies to determine optimal controller gains, ensuring accurate trajectory tracking under non-holonomic constraints. The methodology is validated through simulations of circular and lemniscate-shaped trajectories, where the objective function is defined as the Root Mean Square Error (RMSE) between the desired and actual trajectories. Results show that the proposed approach achieves highly accurate tracking with RMSE values close to 0.028 m in both scenarios, demonstrating WSO's efficiency, convergence stability, and potential as a competitive alternative to classical optimization methods in mobile robotics.
Data analysis of a group of different indexes and machine learning methods are used to obtain current progress and projections of the future achievements of the UN sustainable development goals. These projections can be a valuable tool to make educated decisions that improve a country's performance in seeking to meet the UN sustainable developments goals. This work explores the Sustainable Development Goals Index (SDGI) and its correlations with another seven important indexes: Urban Competence Index (UCI), Fragile States Index (FSI), Human Development Index (HDI), Global Competitiveness Index (GCI), Global Innovation Index (GII), Environmental Performance Index (EPI) and Social Progress Index (SPI). Machine learning projections are obtained for the time evolution of the indexes with the purpose of knowing in what percentage the sustainability objectives set by the 2030 United Nations agenda will be achieved by our study case Mexico. Using Pearson correlations, multiple linear approximations of the SGD index are proposed based on indexes that have a statistically significant correlation with it. Our findings suggest that indexes correlated with de SDGI can be used as predictors, thus a close monitoring of these predictor variables can give us valuable information on how the SDGI will change. Machine learning projections show us the future behavior not only of the SDGI but also of the predictor indexes that are correlated with it. Currently to evaluate progress in achieving the goals, the United Nations periodically reports the measurement of the distance to the SDG targets for various countries. However, oftentimes not enough data is provided for the indicators, which makes it difficult to establish progress towards a certain SDG. Therefore, understanding the multiple linear relationship between the SDGI and other predictive indexes might help gauge the SDGI's advancement.
In domestic environments, most users lack access to detailed, real-time data on their electricity consumption. This limitation hinders their ability to make informed decisions that promote energy efficiency and cost savings. Existing utility tools typically present historical consumption data in static formats, making identifying specific usage patterns or high-consumption devices challenging. Furthermore, the lack of visual and accessible digital interfaces in this area prevents users, especially those without technical knowledge, from understanding their energy consumption. To address this problem, we present Power Track, an interactive, user-centric system designed to monitor, control, and visualize real-time domestic energy consumption. The platform integrates sensors, ESP32 microcontrollers, and Wi-Fi communication via the MQTT protocol to collect data stored in a cloud-based relational database. The system features a responsive web interface, developed with React and Node.js, that graphically displays historical and current consumption. This design fosters energy awareness and facilitates informed decision-making, encouraging sustainable consumption habits. Development followed a User-Centered Design (UCD) methodology, which included requirements specification, prototyping, and user validation to ensure usability and relevance. Power Track represents a scalable and user-friendly tool that aligns with the principles of smart cities and energy efficiency.
This study evaluates the impact of Math Modelation Quest (MMQ), a gamified mobile application designed to enhance algebra learning and promote technology acceptance among high school students. A quasi-experimental design was implemented with 70 participants, combining pre- and post-tests to measure algebra performance and a Technology Acceptance Model (TAM) survey to assess perceptions of the app. After three in-class sessions, results revealed a statistically significant improvement in algebra scores, with the group average increasing by 7 points, benefiting both lower- and higher-achieving students as well as both genders. The overall TAM mean was 3.06 on a $0-4$ Likert scale, indicating favorable acceptance of the app, with perceived ease of use as the strongest dimension. These findings confirm MMQ's potential as a scalable, student-centered tool that fosters both measurable academic progress and positive attitudes toward digital learning and extend previous pilot results with undergraduate students by demonstrating similar benefits for high school learners.
This work presents the design and training of a large language model (LLM) for bidirectional translation between Spanish and Nahuatl, focusing on technical accuracy and cultural relevance. A bilingual corpus of 7,500 sentence pairs was constructed, a Transformer model was trained from scratch with Sentence-Piece tokenization, and a functional web prototype was deployed. BLEU and chrF++ metrics confirm the viability of the system, which preserves linguistic features of Nahuatl frequently omitted by generic translators. The model has immediate applications in educational and community contexts and represents a technical and social contribution toward digital inclusion and the preservation of indigenous languages. The work was carried out at the TecNM/Apizaco Institute of Technology (ITA) and supported by the Secretariat of Science, Humanities, Technology and Innovation (SECIHTI).
Pronosupination movement of the human forearm allows essential activities of daily living such as feeding or grooming. A forearm prosthesis can imitate the biomechanics of pronosupination. However, the design of the forearm prosthesis has significant challenges to enhance its performance. Herein, we design a prototype for a forearm prosthesis with a pronosupination movement based on a simple mechanism. This prototype can generate the motions of the pronosupination of the human forearm with good performance. Furthermore, the proposed prototype includes a simple mechanism, which decreases its manufacturing cost. The prototype's performance is tested with suitable results, reproducing the biomechanics of the human forearm using an easy-to-operate mechanism. This prototype design can be optimized for future research on a human forearm prosthesis.
This work presents the design and implementation of a hybrid prototype that integrates a Wireless Sensor Network (WSN) with an optical link using direct optical modulation. The WSN emulates nodes across crop field areas to monitor parameters such as humidity and temperature. The optical communication system permits real-time transmission of this data over long distances. Precisely, this is the main contribution of this work, as the distance limitation by the use of wireless devices is overcome. The results demonstrate the robustness of the link, bandwidth optimization, and efficiency of centralized data analysis. The proposed solution is scalable and adaptable to remote agricultural environments, contributing to sustainable resource management.
Type 1 Diabetes Mellitus (T1DM) requires precise glucose management to prevent complications. A significant challenge is determining the optimal insulin bolus dose, as errors can lead to hypo- or hyperglycemia. This paper presents an application of Directed Acyclic Graphs (DAGs) to model causal relationships that influence postprandial glucose levels using the HUPA-UCM dataset (25 T1DM patients with glucose, insulin, activity, and dietary data). Three DAG models were generated: knowledge-based (DAG1), data-driven (DAG2), and adjusted (DAG3). Causal inference was performed using DoWhy, revealing a statistically significant effect of insulin bolus on postprandial glucose. Each additional bolus unit reduced glucose between -1.30 mg/dL (DAG1) and -0.78 mg/dL (DAG2/DAG3). Robust tests confirmed stability (p > 0.90). Results demonstrate that DAGs can reliably identify determinants of glycemic response and represent a promising tool for interpretable, personalized insulin dosing systems. These results could support the design of intelligent insulin dosing systems capable of explaining their recommendations to clinicians and patients.
This qualitative case study examines a binational Collaborative Online International Learning (COIL) activity that uses matrix cryptography to teach matrix inversion and connect linear algebra with SDG 4 (Quality Education). Forty undergraduate students from Mexico and Chile worked in mixed teams to decrypt a message encoded with a 10x10 matrix using Python/MATLAB/Excel. They then linked the decrypted content to SDG 4 targets and proposed feasible on-campus actions. Data sources included team reports, recorded presentations, and individual reflections. An inductive, reflexive thematic analysis-supported by Atlas.ti-identified three interrelated dimensions: (i) technical (procedural autonomy, cross-tool validation, and understanding of invertibility conditions), (ii) pedagogical (meaningful learning under the SDG 4 framing), and (iii) emotional-collaborative (motivation, intercultural communication, and sense of purpose). We present qualitative indications of conceptual understanding of invertibility, based on textual evidence of diagnosis and justification in student artifacts; however, no objective pre/post tests were administered. We discuss design implications for STEM courses and propose a replicable rubric for post hoc conceptual evaluation.
The integration of Large Language Models (LLMs) into biomedical research offers new opportunities for domain-specific reasoning and knowledge representation. However, their performance depends heavily on the semantic quality of training data. In oncology, where precision and interpretability are vital, scalable methods for constructing structured knowledge bases are essential for effective fine-tuning. This study presents a pipeline for developing a lung cancer knowledge base using Open Information Extraction (OpenIE). The process includes: (1) identifying medical concepts with the MeSH thesaurus; (2) filtering open-access PubMed literature with permissive licenses (CC0); (3) extracting (subject, relation, object) triplets using OpenIE method; and (4) enriching triplet sets with Named Entity Recognition (NER) to ensure biomedical relevance. The resulting triplet sets provide a domain-specific, large-scale, and noise-aware resource for fine-tuning LLMs. We evaluated T5 models fine-tuned on this dataset through Supervised Semantic Fine-Tuning. Comparative assessments with ROUGE and BERTScore show significantly improved performance and semantic coherence, demonstrating the potential of OpenIE-derived resources as scalable, low-cost solutions for enhancing biomedical NLP.
