Esvan-Jesús Pérez-Pérez

Water-infrastructure digitalization increasingly couples hydraulics with sensing, communication, artificial intelligence (AI), and supervisory control, while digital twins support operational decisions. This article combines a critical narrative review with structured full-text evidence mapping of 50 water cyber-physical systems (CPS) studies and an accompanying reproducible Water Distribution Testbed (WDT) benchmark case study. The review synthesizes model-based, data-driven, and physically informed methods across detection, localization, mitigation, and recovery, distinguishing anomaly evidence, cyber attribution, and closed-loop resilience. The WDT case study compares nominal-only detectors and alarm policies under a common chronological protocol. Across five primary neural runs, cumulative policies increase sample-level recall but sustain alarms and detect fewer event onsets than the fixed-threshold policy. These numerical findings characterize the declared protocol and evaluated records; they do not establish a general ranking of detection algorithms. They motivate interpreting classification metrics alongside event performance and held-out nominal alarm burden. Priorities include trustworthy digital-twin updating, causal localization, uncertainty-aware intervention, independent validation, and verified recovery for intelligent water-infrastructure operation.

Wildfires are events in which fire spreads uncontrollably, destroying natural ecosystems and causing severe biodiversity loss. Driven by climate change, the frequency and intensity of these events have escalated, making effective risk mitigation a critical global priority. This study proposes a hybrid methodology for the identification and classification of wildfire risk zones, based on the combination of a convolutional neural network (CNN) with U-Net architecture and ResNet-50 backbone for semantic segmentation, together with a fuzzy inference system. The CNN processes high-resolution georeferenced RGB imagery to identify environmental patterns such as vegetation density and combustible organic matter. Evaluated through cross-validation, the CNN achieved a global IoU of 86.73% and a global F1-score of 92.82%, generating a risk classification into low, medium, and high levels. Subsequently, this categorical output is integrated into a fuzzy inference system along with meteorological variables (temperature, humidity, and wind speed). The fuzzy inference system dynamically adjusts the initial risk assessment according to meteorological conditions and generates an updated risk classification for the processed images. This approach significantly improves wildfire risk assessment, providing a data-driven tool for environmental management and early disaster prevention.

This paper proposes to estimate the travel time at each edge of an urban street network using Artificial Neural Networks (ANNs). To improve the ANN performance and minimize errors in manual design, an Algorithm Auto-Design ANN Topology (A-DANNT) is introduced that automatically determines the most suitable architecture for regression problems. The methodology implements an algorithm based on Tabu Search, called the Best R-Value Determination algorithm (BR-vD), which optimizes the topology obtaining a lower Mean Square Error (MSE) and a higher correlation coefficient. The process is developed in three phases: first, the variables that impact the travel time are analyzed; then, the proposed algorithm is used to find the best topology; and finally, the travel times are estimated. The proposal is evaluated in two case studies: in the first, the algorithm automatically designs the architecture, and a 0.99366 correlation coefficient is achieved between the results and the objectives. In the second case, the performance of the algorithm is compared with a fuzzy travel time model, achieving a 0.99898 correlation coefficient. In both cases, the proposed algorithm is capable of designing topologies with coefficients greater than 0.99 and Mean Absolute Errors (MAEs) of 3.2765 and 0.6957 s, respectively.

This study presents a low-cost and scalable CO₂ monitoring system that leverages NDIR sensors and a Long Short-Term Memory (LSTM) neural network to predict indoor CO₂ concentrations over both short- and long-term horizons. The proposed system aims to anticipate air quality deterioration in shared spaces, enabling proactive ventilation strategies. Various LSTM configurations were evaluated, optimizing the number of layers, neurons per layer, and input delays to enhance forecasting accuracy. The optimal model consisted of two LSTM layers with 128 neurons each and a time window of 10 previous observations. This model achieved an RMSE of approximately 57 ppm for an 8 h forecast in a classroom setting. Experimental results demonstrate the reliability of the proposed approach for CO₂ prediction and its potential impact on indoor air quality management.

This work presents a data-driven approach for diagnosing sensor faults and leaks in hydraulic pipelines using neuro-fuzzy Zonotopic Kalman Filters (ZKF). The approach involves two key steps: first, identifying the nonlinear pipeline system using an adaptive neuro-fuzzy inference system (ANFIS), resulting in a set of Takagi–Sugeno fuzzy models derived from pressure and flow data, and second, implementing a neuro-fuzzy ZKF bench to detect pipeline leaks and sensor faults with adaptive thresholds. The learning phase of the neuro-fuzzy systems considers only fault-free data. Fault isolation is achieved by comparing zonotopic sets and evaluating a fault signature matrix. The method accounts for parametric uncertainty and measurement noise, ensuring robustness. Experimental validation on a hydraulic pipeline demonstrated high precision (up to 99.24%), recall (up to 99.20%), and low false positive rates (as low as 0.76%) across various fault scenarios and operational points.