Francesc Pozo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Francesc Pozo has a total of 113 co-authors. Here are the top co-authors by number of shared publications:
- Yolanda Vidal
Control, Data, and Artificial Intelligence (CoDAlab), Department of Mathematics, Escola d’Enginyeria de Barcelona Est (EEBE), Universitat Politècnica de Catalunya (UPC), Campus Diagonal-Besós (CDB), Eduard Maristany, 16, 08019 Barcelona, Spain
33 shared publications - José Rodellar
Universitat Politecnica de Catalunya
27 shared publications - Leonardo Acho
Department of Mathematics, Terrassa School of Industrial, Aerospace and Audiovisual Engineering (ESEIAAT), Universitat Politècnica de Catalunya-BarcelonaTech (UPC), 08222 Barcelona, Spain
27 shared publications - Diego Alexander Tibaduiza
Departamento de Ingeniería Eléctrica y Electrónica, Universidad Nacional de Colombia, Bogotá 111321, Colombia
23 shared publications - Jersson Xavier León Medina
Escuela de Ingeniería Electromecánica, Facultad Seccional Duitama, Universidad Pedagógica y Tecnológica de Colombia (UPTC), Carrera 18 con Calle 22, Duitama 150461, Colombia
16 shared publications - Maribel Anaya
Departamento de Ingeniería Eléctrica y Electrónica, Universidad Nacional de Colombia, Bogotá 111321, Colombia
15 shared publications
Latest Publications
- Open Access
Improving Vibration-Based Classification in a Wind Turbine Jacket Structure and Rotor Through Multitask Learning and Spectral Representations
This work proposes and evaluates a supervised classification framework based on multichannel vibration signals to analyze two systems associated with wind turbines: rotor imbalance in a laboratory-adapted Enair E30Pro wind turbine and structural damage in a jacket-type structure. The methodology compares single-task and multitask learning models using spectral representations obtained from the Fast Fourier Transform (FFT) in log-magnitude format, the Short-Time Fourier Transform (STFT) in log-power format, and the Welch power spectral density (Welch PSD) in log-power format. During the single-task stage, TabNet (Attentive Interpretable Tabular Learning) models, a deep learning architecture designed for tabular data, were independently trained for each system and representation to establish baseline performance levels. Subsequently, a multitask scheme composed of a shared encoder and two task-specific classification heads was implemented: one for the jacket-type structure and another for the rotor. In addition to the original TabNet baselines, a capacity-matched single-task MLP control was evaluated using the same encoder architecture, preprocessing, optimization settings, epoch budget, and cross-validation folds as the multitask model. Relative to the TabNet baselines, the multitask model achieved higher performance in several configurations, particularly for the more challenging rotor task; however, the capacity-matched control showed that these gains cannot be attributed uniquely to multitask parameter sharing because the multitask model did not consistently outperform an equivalently structured single-task network. For the jacket-type structure, all three spectral representations achieved near-perfect classification performance, with Macro-F1 values above 0.99 in both the single-task and multitask settings. For the rotor imbalance problem, the multitask model using Welch PSD log-power achieved a global accuracy of 0.9125 and a Macro-F1 of 0.8909. Negative-control experiments further showed that replacing the auxiliary-task labels with random permutations reduced target-task performance in several configurations, indicating that informative auxiliary supervision influences the shared optimization process, although this effect did not translate into a consistent predictive advantage over the capacity-matched single-task control. The multitask architecture nevertheless provides a parameter-efficient joint formulation: one shared multitask network contains 4,481,354 trainable parameters, compared with 8,958,218 parameters required by two independent capacity-matched single-task networks, corresponding to an approximately 49.97% reduction in total model parameters. Overall, the results support multitask learning as a compact joint modeling strategy for physically different vibration-based classification problems, while indicating that its principal benefit in the present study is architectural efficiency rather than a consistently superior predictive effect attributable to positive cross-task transfer.
Open AccessImproving Vibration-Based Classification in a Wind Turbine Jacket Structure and Rotor Through Multitask Learning and Spectral Representations
This work proposes and evaluates a supervised classification framework based on multichannel vibration signals to analyze two systems associated with wind turbines: rotor imbalance in a laboratory-adapted Enair E30Pro wind turbine and structural damage in a jacket-type structure. The methodology compares single-task and multitask learning models using spectral representations obtained from the Fast Fourier Transform (FFT) in log-magnitude format, the Short-Time Fourier Transform (STFT) in log-power format, and the Welch power spectral density (Welch PSD) in log-power format. During the single-task stage, TabNet (Attentive Interpretable Tabular Learning) models, a deep learning architecture designed for tabular data, were independently trained for each system and representation to establish baseline performance levels. Subsequently, a multitask scheme composed of a shared encoder and two task-specific classification heads was implemented: one for the jacket-type structure and another for the rotor. In addition to the original TabNet baselines, a capacity-matched single-task MLP control was evaluated using the same encoder architecture, preprocessing, optimization settings, epoch budget, and cross-validation folds as the multitask model. Relative to the TabNet baselines, the multitask model achieved higher performance in several configurations, particularly for the more challenging rotor task; however, the capacity-matched control showed that these gains cannot be attributed uniquely to multitask parameter sharing because the multitask model did not consistently outperform an equivalently structured single-task network. For the jacket-type structure, all three spectral representations achieved near-perfect classification performance, with Macro-F1 values above 0.99 in both the single-task and multitask settings. For the rotor imbalance problem, the multitask model using Welch PSD log-power achieved a global accuracy of 0.9125 and a Macro-F1 of 0.8909. Negative-control experiments further showed that replacing the auxiliary-task labels with random permutations reduced target-task performance in several configurations, indicating that informative auxiliary supervision influences the shared optimization process, although this effect did not translate into a consistent predictive advantage over the capacity-matched single-task control. The multitask architecture nevertheless provides a parameter-efficient joint formulation: one shared multitask network contains 4,481,354 trainable parameters, compared with 8,958,218 parameters required by two independent capacity-matched single-task networks, corresponding to an approximately 49.97% reduction in total model parameters. Overall, the results support multitask learning as a compact joint modeling strategy for physically different vibration-based classification problems, while indicating that its principal benefit in the present study is architectural efficiency rather than a consistently superior predictive effect attributable to positive cross-task transfer.
- Open Access
- 4 May 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (5) , 363
A Reproducible Benchmarking Methodology for Machine Learning Hardware: Performance–Energy Trade-Offs from GPUs to Apple Silicon
While hardware selection is widely recognized as a key factor in machine learning performance, systematic and reproducible evaluation across heterogeneous and accessible platforms remains limited, particularly when jointly considering execution time, energy consumption, stability, and cost-efficiency. This work presents a unified and fully reproducible benchmarking framework for supervised learning, designed to enable controlled and comparable evaluation across diverse hardware environments. The proposed methodology enforces consistent training pipelines, fixed hyperparameter configurations, and repeated executions to ensure statistical reliability, while incorporating performance metrics such as execution time, power consumption, and energy usage, as well as performance-per-dollar. The framework is validated on a representative set of platforms, including CUDA-enabled GPUs, Apple Silicon (CPU/GPU), x86 processors, ARM-based embedded systems, and cloud-based environments, using convolutional, recurrent (RNN, LSTM, BiLSTM), and tree-based (XGBoost) models. The results reveal that hardware efficiency is strongly model-dependent. GPUs provide the highest computational performance and stability for parallel workloads, whereas Apple Silicon achieves superior energy efficiency with competitive execution times, particularly for recurrent architectures. The batch size analysis shows that performance can vary significantly depending on workload configuration, especially on CPU-based platforms, while epoch-based evaluation confirms that the measured performance reflects steady-state behavior rather than initialization overhead. In contrast, conventional CPUs and embedded systems exhibit significant scalability limitations for deep learning training, although they remain competitive for tree-based methods such as XGBoost, which demonstrates near hardware-independent predictive performance. These findings highlight the limitations of generalized hardware selection criteria and emphasize the need for model-aware and hardware-aware benchmarking. The proposed framework offers a practical and extensible foundation for reproducible, hardware-aware evaluation of machine learning systems, supporting informed decision-making in research, deployment, and cost-constrained scenarios.
- 4 May 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (5) , 363
Open AccessA Reproducible Benchmarking Methodology for Machine Learning Hardware: Performance–Energy Trade-Offs from GPUs to Apple Silicon
While hardware selection is widely recognized as a key factor in machine learning performance, systematic and reproducible evaluation across heterogeneous and accessible platforms remains limited, particularly when jointly considering execution time, energy consumption, stability, and cost-efficiency. This work presents a unified and fully reproducible benchmarking framework for supervised learning, designed to enable controlled and comparable evaluation across diverse hardware environments. The proposed methodology enforces consistent training pipelines, fixed hyperparameter configurations, and repeated executions to ensure statistical reliability, while incorporating performance metrics such as execution time, power consumption, and energy usage, as well as performance-per-dollar. The framework is validated on a representative set of platforms, including CUDA-enabled GPUs, Apple Silicon (CPU/GPU), x86 processors, ARM-based embedded systems, and cloud-based environments, using convolutional, recurrent (RNN, LSTM, BiLSTM), and tree-based (XGBoost) models. The results reveal that hardware efficiency is strongly model-dependent. GPUs provide the highest computational performance and stability for parallel workloads, whereas Apple Silicon achieves superior energy efficiency with competitive execution times, particularly for recurrent architectures. The batch size analysis shows that performance can vary significantly depending on workload configuration, especially on CPU-based platforms, while epoch-based evaluation confirms that the measured performance reflects steady-state behavior rather than initialization overhead. In contrast, conventional CPUs and embedded systems exhibit significant scalability limitations for deep learning training, although they remain competitive for tree-based methods such as XGBoost, which demonstrates near hardware-independent predictive performance. These findings highlight the limitations of generalized hardware selection criteria and emphasize the need for model-aware and hardware-aware benchmarking. The proposed framework offers a practical and extensible foundation for reproducible, hardware-aware evaluation of machine learning systems, supporting informed decision-making in research, deployment, and cost-constrained scenarios.
- Open Access
A Reinforcement Learning-Based Framework for Tariff-Aware Load Shifting in Energy-Intensive Manufacturing
Optimizing energy-intensive manufacturing under time-varying electricity tariffs requires scheduling strategies that reduce cost without compromising operational feasibility. This study is grounded in readily available industrial sensing: we exclusively use time-series measurements of aggregated active power and energy at the main distribution board of a quicklime production plant. We propose a tariff-aware load-shifting framework in which a Proximal Policy Optimization (PPO) reinforcement learning agent is trained in a custom Gymnasium environment to apply discrete consumption scaling actions constrained to 80–125% of a baseline profile during the operating shift (08:00–16:00), explicitly accounting for demand-charge exposure in the TOU peak window (13:00–15:00). The reward design combines instantaneous electricity cost with cumulative energy-tracking penalties and terms associated with operational constraints. Multi-day validation over working days shows consistent economic benefits, with a median total cost reduction on the order of 10% (narrow IQR) driven by reduced peak-window energy and demand peaks. However, the script-based binary compliance indicators (viol_energy, viol_prod_min) reveal deviations from the energy-balance criterion and occasional minimum-production shortfalls under the tolerances used, highlighting the cost–production trade-off and the need for stricter constraint handling for industrial deployment. In addition, we benchmark against dynamic programming (DP), an alternative RL policy (DQN), and a greedy heuristic (GREEDY), comparing cost; operational performance; and, when applicable, computational efficiency, which positions PPO as a competitive alternative among the considered methods. Overall, this work demonstrates how learning-based decision making can be coupled with real-world industrial sensing infrastructures, providing a data-driven tariff-aware scheduling layer for industrial energy management under practical constraints.Open AccessA Reinforcement Learning-Based Framework for Tariff-Aware Load Shifting in Energy-Intensive Manufacturing
Optimizing energy-intensive manufacturing under time-varying electricity tariffs requires scheduling strategies that reduce cost without compromising operational feasibility. This study is grounded in readily available industrial sensing: we exclusively use time-series measurements of aggregated active power and energy at the main distribution board of a quicklime production plant. We propose a tariff-aware load-shifting framework in which a Proximal Policy Optimization (PPO) reinforcement learning agent is trained in a custom Gymnasium environment to apply discrete consumption scaling actions constrained to 80–125% of a baseline profile during the operating shift (08:00–16:00), explicitly accounting for demand-charge exposure in the TOU peak window (13:00–15:00). The reward design combines instantaneous electricity cost with cumulative energy-tracking penalties and terms associated with operational constraints. Multi-day validation over working days shows consistent economic benefits, with a median total cost reduction on the order of 10% (narrow IQR) driven by reduced peak-window energy and demand peaks. However, the script-based binary compliance indicators (viol_energy, viol_prod_min) reveal deviations from the energy-balance criterion and occasional minimum-production shortfalls under the tolerances used, highlighting the cost–production trade-off and the need for stricter constraint handling for industrial deployment. In addition, we benchmark against dynamic programming (DP), an alternative RL policy (DQN), and a greedy heuristic (GREEDY), comparing cost; operational performance; and, when applicable, computational efficiency, which positions PPO as a competitive alternative among the considered methods. Overall, this work demonstrates how learning-based decision making can be coupled with real-world industrial sensing infrastructures, providing a data-driven tariff-aware scheduling layer for industrial energy management under practical constraints. - Open Access
- 10 March 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (3) , 208
Electric Load Forecasting for a Quicklime Company Using a Temporal Fusion Transformer
Accurate short-term electric load forecasting is essential for the operation and management of energy-intensive manufacturing processes such as quicklime production, for which power demand is driven by stage-based operation, fixed schedules, and abrupt load transitions. This study presents a data-driven forecasting framework based on a Temporal Fusion Transformer (TFT) model applied to real industrial measurements collected during 2024 from an operating quicklime production plant. The dataset comprises hourly average power demand records (kW) measured at a plant level, stage-dependent motor operation, and a fixed working schedule from 08:00 to 18:00 (Monday to Friday), with weekends and non-operational hours characterized by near-zero load. Coke consumption during the calcination stage is included as an additional contextual variable. The TFT model is trained for multi-horizon forecasting and provides probabilistic prediction intervals through quantile regression. Weekly evaluations demonstrate that the proposed approach accurately captures start–stop behavior, peak-load periods, and structured inactivity intervals. In addition to point-wise accuracy metrics, cumulative energy is evaluated by integrating hourly power over the forecasting horizon, allowing the assessment of energy preservation at the operational level. The resulting energy deviation reaches 4.78% for the full horizon and 5.25% when restricted to active production hours, confirming strong consistency between predicted and actual cumulative energy. A comparative analysis against LSTM, GRU, and N-BEATS models shows that recurrent architectures achieve lower MAE and RMSE values, while the TFT model delivers superior cumulative energy consistency, highlighting a trade-off between instantaneous accuracy and operational energy fidelity. Overall, the results demonstrate that the proposed TFT-based framework provides a robust and practically relevant solution for short-term industrial electric load forecasting and decision support in stage-driven manufacturing systems under real operating conditions.
- 10 March 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (3) , 208
Open AccessElectric Load Forecasting for a Quicklime Company Using a Temporal Fusion Transformer
Accurate short-term electric load forecasting is essential for the operation and management of energy-intensive manufacturing processes such as quicklime production, for which power demand is driven by stage-based operation, fixed schedules, and abrupt load transitions. This study presents a data-driven forecasting framework based on a Temporal Fusion Transformer (TFT) model applied to real industrial measurements collected during 2024 from an operating quicklime production plant. The dataset comprises hourly average power demand records (kW) measured at a plant level, stage-dependent motor operation, and a fixed working schedule from 08:00 to 18:00 (Monday to Friday), with weekends and non-operational hours characterized by near-zero load. Coke consumption during the calcination stage is included as an additional contextual variable. The TFT model is trained for multi-horizon forecasting and provides probabilistic prediction intervals through quantile regression. Weekly evaluations demonstrate that the proposed approach accurately captures start–stop behavior, peak-load periods, and structured inactivity intervals. In addition to point-wise accuracy metrics, cumulative energy is evaluated by integrating hourly power over the forecasting horizon, allowing the assessment of energy preservation at the operational level. The resulting energy deviation reaches 4.78% for the full horizon and 5.25% when restricted to active production hours, confirming strong consistency between predicted and actual cumulative energy. A comparative analysis against LSTM, GRU, and N-BEATS models shows that recurrent architectures achieve lower MAE and RMSE values, while the TFT model delivers superior cumulative energy consistency, highlighting a trade-off between instantaneous accuracy and operational energy fidelity. Overall, the results demonstrate that the proposed TFT-based framework provides a robust and practically relevant solution for short-term industrial electric load forecasting and decision support in stage-driven manufacturing systems under real operating conditions.
- Open Access
- 30 January 2026
- Published by Wiley in Structural Control and Health Monitoring
- Vol. 2026 (1)
Enhancing Wind Turbine Diagnostics With SCADA‐Vibration Fusion, Contrastive Learning, and Linear Predictive Coefficients
Wind energy plays a pivotal role in the transition to sustainable power generation. However, maintaining the reliability and efficiency of wind turbine (WT) remains a significant challenge due to complex operational conditions and the high cost associated with unexpected failures. Effective condition monitoring (CM) and predictive maintenance (PM) strategies are critical to mitigate these risks. This study presents a data‐driven fault detection framework that fuses supervisory control and data acquisition (SCADA) data with high‐frequency vibration signals using deep learning techniques to enhance diagnostic performance. Unlike conventional normal behavior models that rely exclusively on healthy data for training, the proposed framework incorporates limited labeled fault data when available. As only a few types of faults and a few samples are typically available in real‐world scenarios, the approach does not assume a complete representation of all possible fault conditions. Instead, it is designed to generalize beyond the specific faults seen during training. This is demonstrated by training the model on healthy conditions and only two known fault types (with labeled data available) and testing it on a third, previously unseen fault type. In particular, Siamese networks with contrastive and reconstruction learning are employed to improve feature representation and anomaly detection. Two distinct methodologies are compared: the first utilizes a binary cross‐entropy (BCE) loss function to classify the healthy or faulty status of the WT, while the second uses a triplet loss function for multiclass representation learning. Both methodologies generate low‐dimensional representations of the input features, also known as embeddings. The resulting feature embeddings are passed through a k ‐means clustering algorithm to improve fault separation and identification. Statistical features are extracted from SCADA data to capture key trends or event information, while the linear prediction coefficient (LPC) method, which models a signal by predicting future values based on its past samples, is applied to the vibration data for better fault characterization. The proposed approach is evaluated using the publicly available ETH Zurich dataset from an Aventa AV‐7 turbine. Experimental results indicate that the fusion of SCADA and vibration‐based diagnostics, in combination with contrastive and representation learning, substantially improves the predictive accuracy and generalization of fault detection models.
- 30 January 2026
- Published by Wiley in Structural Control and Health Monitoring
- Vol. 2026 (1)
Open AccessEnhancing Wind Turbine Diagnostics With SCADA‐Vibration Fusion, Contrastive Learning, and Linear Predictive Coefficients
Wind energy plays a pivotal role in the transition to sustainable power generation. However, maintaining the reliability and efficiency of wind turbine (WT) remains a significant challenge due to complex operational conditions and the high cost associated with unexpected failures. Effective condition monitoring (CM) and predictive maintenance (PM) strategies are critical to mitigate these risks. This study presents a data‐driven fault detection framework that fuses supervisory control and data acquisition (SCADA) data with high‐frequency vibration signals using deep learning techniques to enhance diagnostic performance. Unlike conventional normal behavior models that rely exclusively on healthy data for training, the proposed framework incorporates limited labeled fault data when available. As only a few types of faults and a few samples are typically available in real‐world scenarios, the approach does not assume a complete representation of all possible fault conditions. Instead, it is designed to generalize beyond the specific faults seen during training. This is demonstrated by training the model on healthy conditions and only two known fault types (with labeled data available) and testing it on a third, previously unseen fault type. In particular, Siamese networks with contrastive and reconstruction learning are employed to improve feature representation and anomaly detection. Two distinct methodologies are compared: the first utilizes a binary cross‐entropy (BCE) loss function to classify the healthy or faulty status of the WT, while the second uses a triplet loss function for multiclass representation learning. Both methodologies generate low‐dimensional representations of the input features, also known as embeddings. The resulting feature embeddings are passed through a k ‐means clustering algorithm to improve fault separation and identification. Statistical features are extracted from SCADA data to capture key trends or event information, while the linear prediction coefficient (LPC) method, which models a signal by predicting future values based on its past samples, is applied to the vibration data for better fault characterization. The proposed approach is evaluated using the publicly available ETH Zurich dataset from an Aventa AV‐7 turbine. Experimental results indicate that the fusion of SCADA and vibration‐based diagnostics, in combination with contrastive and representation learning, substantially improves the predictive accuracy and generalization of fault detection models.
- Open Access
Forecasting Energy Demand in Quicklime Manufacturing: A Data-Driven Approach
This study presents a deep learning-based framework for forecasting energy demand in a quicklime production company, aiming to enhance operational efficiency and enable data-driven decision-making for industrial scalability. Using one year of real electricity consumption data, the methodology integrates temporal and operational variables—such as load profile, active power, shift indicators, and production-related proxies—to capture the dynamics of energy usage throughout the manufacturing process. Several neural network architectures, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Conv1D models, were trained and compared to predict short-term power demand with 10-min resolution. Among these, the GRU model achieved the highest predictive accuracy, with a best performance of RMSE = 2.18 kW, MAE = 0.49 kW, and SMAPE = 3.64% on the test set. The resulting forecasts support cost-efficient scheduling under time-of-use tariffs and provide valuable insights for infrastructure planning, capacity management, and sustainability optimization in energy-intensive industries.
Open AccessForecasting Energy Demand in Quicklime Manufacturing: A Data-Driven Approach
This study presents a deep learning-based framework for forecasting energy demand in a quicklime production company, aiming to enhance operational efficiency and enable data-driven decision-making for industrial scalability. Using one year of real electricity consumption data, the methodology integrates temporal and operational variables—such as load profile, active power, shift indicators, and production-related proxies—to capture the dynamics of energy usage throughout the manufacturing process. Several neural network architectures, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Conv1D models, were trained and compared to predict short-term power demand with 10-min resolution. Among these, the GRU model achieved the highest predictive accuracy, with a best performance of RMSE = 2.18 kW, MAE = 0.49 kW, and SMAPE = 3.64% on the test set. The resulting forecasts support cost-efficient scheduling under time-of-use tariffs and provide valuable insights for infrastructure planning, capacity management, and sustainability optimization in energy-intensive industries.
- Open Access
- 24 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 200683-200695
Experimental Analysis of CFRP and GFRP Composites Using Guided Ultrasonic Waves: Development of Frequency Maps
The in situ and non-destructive inspection of composite materials, particularly carbon fiber-reinforced polymers (CFRP) and glass fiber-reinforced polymers (GFRP), is essential in lightweight structural applications where early damage detection is critical for ensuring structural reliability. Guided ultrasonic waves (GUWs) offer a promising non-destructive evaluation (NDE) technique due to their ability to propagate throughout long distances while being sensitive to material defects. This work presents the characterization of experimental GUW propagation features in CFRP and GFRP laminates, providing a comparative assessment of wave behavior in these materials. Experiments were conducted using piezoelectric transducers to generate and receive guided ultrasonic waves under varying excitation frequencies, plate thickness, and transducer arrangement. Key parameters, such as phase and group velocities, and attenuation, were analyzed to identify the impact of fiber type, layup orientation, and internal structural differences on wave propagation. The results indicate that CFRP exhibits lower attenuation and more substantial mode dispersion than GFRP, attributed to its higher stiffness and anisotropy. As a result, CFRP shows better wave transmission with lower attenuation, allowing long-range ultrasonic inspection. This study also provides valuable insights for developing ultrasonic inspection methodologies based on different fiber orientations and multiple laminated composite materials (CFRP and GFRP).
- 24 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 200683-200695
Open AccessExperimental Analysis of CFRP and GFRP Composites Using Guided Ultrasonic Waves: Development of Frequency Maps
The in situ and non-destructive inspection of composite materials, particularly carbon fiber-reinforced polymers (CFRP) and glass fiber-reinforced polymers (GFRP), is essential in lightweight structural applications where early damage detection is critical for ensuring structural reliability. Guided ultrasonic waves (GUWs) offer a promising non-destructive evaluation (NDE) technique due to their ability to propagate throughout long distances while being sensitive to material defects. This work presents the characterization of experimental GUW propagation features in CFRP and GFRP laminates, providing a comparative assessment of wave behavior in these materials. Experiments were conducted using piezoelectric transducers to generate and receive guided ultrasonic waves under varying excitation frequencies, plate thickness, and transducer arrangement. Key parameters, such as phase and group velocities, and attenuation, were analyzed to identify the impact of fiber type, layup orientation, and internal structural differences on wave propagation. The results indicate that CFRP exhibits lower attenuation and more substantial mode dispersion than GFRP, attributed to its higher stiffness and anisotropy. As a result, CFRP shows better wave transmission with lower attenuation, allowing long-range ultrasonic inspection. This study also provides valuable insights for developing ultrasonic inspection methodologies based on different fiber orientations and multiple laminated composite materials (CFRP and GFRP).
- Top Cited PapersOpen Access
- 30 October 2025
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 267
Recent advances in wind turbine condition monitoring using SCADA data: A state-of-the-art review
- 30 October 2025
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 267
Top Cited PapersOpen AccessRecent advances in wind turbine condition monitoring using SCADA data: A state-of-the-art review
- 1 October 2025
- Published by Springer Nature
- p. 962-969
Wind Turbine Gearbox Fault Detection Under Distribution Shifts
- 1 October 2025
- Published by Springer Nature
- p. 962-969
Wind Turbine Gearbox Fault Detection Under Distribution Shifts
- Top Cited PapersOpen Access
- 28 May 2025
- Published by Elsevier in Intelligent Systems with Applications
- Vol. 26
Digital twin technology in wind turbine components: A review
- 28 May 2025
- Published by Elsevier in Intelligent Systems with Applications
- Vol. 26
Top Cited PapersOpen AccessDigital twin technology in wind turbine components: A review