Fernando A. Inthamoussou | Instituto de Investigaciones en Electrónica, Control y Procesamiento de Señales

Formación Académica

Ingeniero en Electrónica, Facultad de Ingeniería, UNLP.
Doctor en Ingeniería, Facultad de Ingeniería, UNLP.

Posición actual

Profesor Adjunto, Facultad de Ingeniería, UNLP.
Investigador Asistente, CONICET, Argentina.

Área de Investigación actual

Perfiles

Página personal

2025

M. Saavedra; N. Faedo; F. Inthamoussou; F. Mosquera; F. Garelli

Comparative evaluation of data-based estimators for wave-induced force in wave energy converters Artículo de revista

En: J. Ocean Eng. Mar. Energy, 2025, ISSN: 2198-6452.

Resumen | Enlaces | BibTeX

@article{Saavedra2025,

title = {Comparative evaluation of data-based estimators for wave-induced force in wave energy converters},

author = {M. Saavedra and N. Faedo and F. Inthamoussou and F. Mosquera and F. Garelli},

doi = {10.1007/s40722-025-00427-4},

issn = {2198-6452},

year  = {2025},

date = {2025-09-01},

urldate = {2025-09-01},

journal = {J. Ocean Eng. Mar. Energy},

publisher = {Springer Science and Business Media LLC},

abstract = {<jats:title>Abstract</jats:title>

          <jats:p>Wave energy conversion technology emerges as a promising approach to renewable energy generation, offering a consistent and predictable power source that complements intermittent renewable energy sources such as solar and wind power. Achieving optimal ocean wave energy absorption requires precise knowledge of the so-called wave excitation force, which is typically estimated through model-based techniques reliant on accurate system descriptions. However, uncertainties inherent to hydrodynamic modelling often limit the reliability of these approaches. To address this challenge, this paper presents a comprehensive evaluation of model-free data-based estimators, for wave excitation torque estimation in Wavestar like wave energy converters (WECs). The study examines various neural network architectures, including static models (feedforward networks) and those incorporating temporal dynamics (recurrent neural networks and long short-term memory networks). The analysis examines the impact of utilising multiple input combinations, ranging from motion variables to configurations enhanced with surrounding wave height measurements from the device’s vicinity. Input selection is guided by correlation analysis and spectral coherence evaluation to ensure physical relevance and practical feasibility. Estimators are trained and tested using experimental data obtained from a comprehensive wave tank campaign emulating diverse sea state conditions. The results demonstrate that architectures incorporating temporal considerations achieve superior performance, particularly under wide-banded sea states. A comparative analysis with a model-based estimator, implemented via a Kalman–Bucy Filter with a harmonic oscillator expansion, highlights the advantages of neural networks, especially under challenging conditions where model-based approaches face significant limitations. These findings underscore the capability of data-based strategies to reduce dependence on potentially complex and uncertain analytical models, offering a promising alternative for improving WEC control systems.</jats:p>},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

Cerrar

Abstract
Wave energy conversion technology emerges as a promising approach to renewable energy generation, offering a consistent and predictable power source that complements intermittent renewable energy sources such as solar and wind power. Achieving optimal ocean wave energy absorption requires precise knowledge of the so-called wave excitation force, which is typically estimated through model-based techniques reliant on accurate system descriptions. However, uncertainties inherent to hydrodynamic modelling often limit the reliability of these approaches. To address this challenge, this paper presents a comprehensive evaluation of model-free data-based estimators, for wave excitation torque estimation in Wavestar like wave energy converters (WECs). The study examines various neural network architectures, including static models (feedforward networks) and those incorporating temporal dynamics (recurrent neural networks and long short-term memory networks). The analysis examines the impact of utilising multiple input combinations, ranging from motion variables to configurations enhanced with surrounding wave height measurements from the device’s vicinity. Input selection is guided by correlation analysis and spectral coherence evaluation to ensure physical relevance and practical feasibility. Estimators are trained and tested using experimental data obtained from a comprehensive wave tank campaign emulating diverse sea state conditions. The results demonstrate that architectures incorporating temporal considerations achieve superior performance, particularly under wide-banded sea states. A comparative analysis with a model-based estimator, implemented via a Kalman–Bucy Filter with a harmonic oscillator expansion, highlights the advantages of neural networks, especially under challenging conditions where model-based approaches face significant limitations. These findings underscore the capability of data-based strategies to reduce dependence on potentially complex and uncertain analytical models, offering a promising alternative for improving WEC control systems.

Cerrar

M. Saavedra; F. Inthamoussou; E. Fushimi; F. Garelli

Identification of Physical Activity Type in People with Diabetes: A Spectrogram-Based Approach Artículo de revista

En: Diabetes Technology and Obesity Medicine, vol. 1, no 1, pp. 361-373, 2025.

Resumen | Enlaces | BibTeX

@article{doi:10.1177/29941520251358842,

title = {Identification of Physical Activity Type in People with Diabetes: A Spectrogram-Based Approach},

author = { M. Saavedra and F. Inthamoussou and E. Fushimi and F. Garelli},

url = {https://www.liebertpub.com/doi/abs/10.1177/29941520251358842},

doi = {10.1177/29941520251358842},

year  = {2025},

date = {2025-07-21},

urldate = {2025-01-01},

journal = {Diabetes Technology and Obesity Medicine},

volume = {1},

number = {1},

pages = {361-373},

abstract = {Background: Individuals with type 1 diabetes (T1D) require close glucose monitoring to prevent both short- and long-term complications. Physical activity (PA) is a significant source of variability in metabolic dynamics, leading to glycemic fluctuations that depend on the type, intensity, and duration of the exercise. Accurately monitoring and classifying the type of PA is crucial for optimizing glycemic control and minimizing the risk of hypoglycemia. Method: This study utilizes the largest clinical trial of PA in people with T1D to date, the Type 1 Diabetes and Exercise Initiative (T1DEXI), which included both structured and unstructured PA sessions, to develop an online classification approach for identifying the type of PA (aerobic, interval, resistance). A computationally efficient convolutional neural network (CNN) was trained on time–frequency representations (spectrograms) of step count and heart rate signals, readily available from wearable devices, from the structured PA sessions of the T1DEXI dataset. The proposed methodology presents an ad hoc process for designing the spectrograms based on the CNN architecture to optimize the classifier’s performance. Results: The CNN-based classification approach was implemented using spectrograms of 5- and 30-min signals, resulting in two classifiers that achieve high classification accuracy when evaluated on the structured PA sessions. The 5-min classifier was then applied to unstructured PA sessions, where the predicted distribution of glucose changes for the activity types was consistent with clinical evidence. Conclusion: These results demonstrate the potential of the proposed approach for its integration into decision support systems or automated insulin delivery systems, enabling improved glucose management during exercise in T1D.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}