Understanding Fabrication Variability in Core-Shell Soft Biomaterials Using Stochastic Artificial Intelligence
Approaches for the fabrication of biomaterials are currently numerous, with a wide diversity of available material precursors, chemistries, and processing technologies. Owing to the complex nature of the human body, biomaterials are targeted for applications with highly diverse performance demands. Traditional strategies based on trial and error have fallen short of predicting the ideal parameters required to produce adequate structures to meet these challenges. Although the design of experiments enables reducing experimental testing, it has failed to predict complex, multi-factorial processing effects, including experimental variability. Despite being often overlooked, experimental variability is an important aspect in biomaterials, which are often processed from source materials with significant compositional variability (e.g., natural polymers), along with processing methodologies frequently undertaken under poorly controlled environmental conditions. Here, a machine learning approach based on Gaussian processes (GPs) is developed to identify patterns and correlations between fabrication conditions and material properties. Flexible soft membrane-based tubular materials obtained by polyelectrolyte complexation are used as a model biomaterial characterized by multi-parametric design inputs. Using GPs, the effects of processing parameters on the magnitude and variability of key properties like permeability, porosity, thickness, opacity, and swelling ratio are quantified. This approach is expected to enable more reliable and predictable biomaterial fabrication.
High-Resolution Distributed Optical Fiber Sensor for Multiparameter Sensing, Local Inhomogeneity Detection, and Poisson’s Ratio Estimation in Concrete Structures
The structural health monitoring (SHM) concept emerges as an important approach for the continuous assessment of civil infrastructure, directly aligning with the demands for sustainability, safety, and smart devices in modern urbanization and smart cities. Addressing these demands, this article presents the development and validation of a multiparameter distributed optical fiber sensor (DOFS) for the simultaneous measurement of temperature, longitudinal strain, and transverse strain with subcentimeter-scale spatial resolution. Based on optical frequency domain reflectometry (OFDR), the system utilizes a geometrically configured commercial fiber to act as a continuous spatial rosette. The sensor was characterized, achieving linear sensitivities of 0.1253 GHz/ μϵ and 1.251 GHz/°C, with a limit of detection of 0.1 μϵ and 0.1 ∘ C and a spatial resolution of 2.5 mm. Application tests on concrete cylinders under compressive loading demonstrated the system’s capability to decouple multiaxial strain components while inherently compensating for thermal cross-sensitivity, maintaining a stable temperature variation of 0.3 ∘ C during testing. The decomposed experimental strain profiles were validated against a numerical finite element analysis (FEA). In contrast with the idealized numerical model, the distributed sensor successfully mapped real-world localized strain concentrations and inherent material inhomogeneities, registering spatial strain variations of up to 619.22 μϵ (longitudinal) and 113.75 μϵ (transverse) near the elastic limit (160 kN). By aggregating the synchronized multitest data, the Poisson’s ratio of the concrete was estimated at 0.16. Furthermore, the system captured the complete spatiotemporal evolution of damage, precisely identifying the macroscopic rupture region. Finally, the results demonstrate that the proposed device provides a robust, single-fiber solution for 2-D strain field mapping and multiparameter SHM, filling key technological gaps for future smart infrastructure applications.
Numerical investigation of a Sagnac-based interferometer for liquid analyte sensing in the terahertz gap
Real-time chemical sensing is critical for industrial quality control and safety. However, most terahertz (THz) photonic crystal fiber (PCF) sensors remain limited to material characterization rather than practical transductors. Here, we introduce Sagnac interferometry to the terahertz domain as a transduction platform for sensing, demonstrating a numerical proof-of-concept liquid sensor using a highly birefringent porous-core photonic crystal fiber infiltrated with an ethanol–water mixture analyte as a proof-of-concept. By translating birefringence modulation into spectral shifts, our simulations show that the sensor achieves sensitivities up to 549 MHz per % v/v, with excellent polynomial correlation ( ) across a wide concentration range, when operating in the 0.56–0.70 THz band. This work establishes a versatile and straightforward interferometric platform in the THz gap, highlighting the potential of this interferometric approach for future chemical and biological applications in the terahertz range.
Low-Cost FBG Interrogator Using Random Nanoparticle-Doped Fiber Laser and Fiber Optic Interferometer
The use of fiber Bragg gratings (FBGs) is widespread in a broad range of applications. However, one of the main drawbacks is the cost and practicality of FBG interrogation systems, which are generally bulky and more expensive than electronic sensors or alternative optical fiber sensing approaches. This article presents a novel, low-cost FBG interrogator that employs a random laser based on a nanoparticle-doped fiber (RNPFL) and an unbalanced Mach–Zehnder fiber interferometer (MZI). The system uses the RNPFL connected to an unbalanced interferometer to convert Bragg wavelength shifts into phase changes, enabling precise interferometric detection with higher resolution than conventional spectral methods. Experimental results obtained with four FBGs of varying reflectivity, spectral width, and sidelobe suppression ratio (SLSR) demonstrate that, when operated with the RNPFL, the interrogator achieves up to 30 dB higher optical power (OP) and interference visibility of 60%–90%, substantially outperforming a standard superluminescent light-emitting diode (SLED) source. The relative error (RE) in phase demodulation remains below 0.2% for optimal configurations, characterizing the precision and phase-noise performance of the demodulation rather than the absolute accuracy of the system. The method is effective for FBGs with a wide range of parameters, provided that the SLSR is at least 15 dB. The main advantages of this approach are its compatibility with standard FBGs, high measurement resolution, and reduced cost, while the primary limitations are the requirement for a star topology in multichannel setups and the sensitivity of the resonator to external disturbances.
Polymer optical fiber Bragg gratings for multiparameter analysis in UAV wings: an AI-integrated approach
This paper presents the development of a polymer optical fiber Bragg grating (POFBG)-based sensor system for measurement of vibration, force amplitude and position for fixed-wing unmanned aerial vehicle (UAV). The POFBG is positioned along the wing structure, where forces amplitude and locations resulted in spectral changes in the POFBG. Then, predefined spectral features are used as input of a Generative Adversarial Network (GAN) for data augmentation, which increased the dataset to 1000 different force amplitude and locations. The accuracy of 1.9 Hz was obtained for vibration analysis. In the force amplitude and location analysis, the feed forward neural network (FFNN) resulted in relative errors of 3.30% and 3.51% for force amplitude and position, such errors are related to root mean squared errors (RMSE) and standard deviations of 0.33 N±0.05 N and 35.18 mm±2.3 mm for force amplitude and location, respectively. If the number of input features is reduced to only 3 predefined spectral features the RMSE without GAN increases to 1.23 N ± 0.15 N and 75.34 mm ± 9.20 mm for force amplitude and position, respectively. The FFNN with data augmentation resulted in relative errors of 7.8% for force amplitude and 2.4% for force location (RMSE of 0.78 N±0.10 N and 24.90 mm±2.1 mm, respectively), using 3 input features. The data augmentation resulted in accuracy increase with almost twofold reduction in the RMSE for the force amplitude and more than 3 times reduction of the RMSE for position estimation. Therefore, the POFBG sensor is a feasible option for shape reconstruction in UAV structures.
Distributed sensing with nanoparticle-doped fibers in standard OTDR systems: Validation and performance analysis
This paper presents the implementation of enhanced backscattering optical fibers in optical time-domain reflectometry (OTDR). The increase in the backscattering is achieved by doping the fiber core with nanoparticles, resulting in the so-called nanoparticle-doped optical fibers (NPFs). The so-called NPF-OTDR sensor system takes advantage of the higher scattering of the NPFs to increase the spatial resolution and strain sensitivity of the distributed sensor system. The wavelength-dependent optical attenuation is measured over a wide range of wavelengths using the cutback method to evaluate the operational wavelength for the OTDR. In this case, the results indicated smaller attenuation in 1310 nm than at the other operational wavelengths of the OTDR system (i.e., 1550 nm and 1625 nm). For this reason, the OTDR is applied at the 1310 nm wavelength for different strains applied along the optical fiber (with 1 m separation between them). The results for strain estimation showed relative errors of around 3%, whereas the strain position estimation is around 0.24 m. These results indicate the feasibility of the proposed system with a potential spatial resolution of around 0.5 m (which can be even lower depending on the OTDR setup). Thus, the advantages of the OTDR using the NPF result in a better sensor system performance when compared with the standard optical fibers, significantly enhancing the conventional OTDR capabilities in terms of resolution with even comparable performance to specialty OTDR systems at a lower cost and with a simpler design.
A Comparison Between Nanoparticle-Doped Fibers and Standard Single-Mode Fibers for Distributed Temperature Sensing
This article presents a comprehensive analysis of the thermal properties and distributed temperature-sensing responses of nanoparticles (NPs)-doped optical fiber using optical frequency-domain reflectometry (OFDR). The thermomechanical analysis (TMA) for the NP-doped fiber is performed to evaluate the thermal expansion coefficient of such fibers and their differences compared with standard telecom optical fibers (SMFs). In addition, the thermo-optical coefficient of such fibers is analyzed as a function of their steady-state temperature responses, whereas the thermal conductivity is assessed from the transient thermal analysis of the NP-doped fiber. The results in distributed temperature measurement indicate better sensing performance of NP-doped fiber, which has 2.6 times higher temperature sensitivity, a smaller response time, and high accuracy estimated from a determination coefficient (R2) of 0.993 and root mean squared error (RMSE) of 1.28 ∘ C. Therefore, the main findings of this work indicate the influence of the NPs composition on the thermal properties of the optical fibers, where the addition of oxide particles (in conjunction with the changes in the glass matrix) presented a positive effect on the distributed temperature sensing capabilities with a thermal expansion coefficient increase of 30% and thermal conductivity increase of 25% when compared with SMF. Thus, this article provides not only the comparison of the temperature sensing performance of an NP-doped fiber but also a straightforward method for the assessment of distributed physical parameters of the fiber that can be used for any specialty optical fiber, which can also be used as a nondestructive method for the assessment of the optical fiber composition.
Fotovoltaico e Aquaponia: Uma Proposta de Tecnologia para a Segurança Alimentar (ODS2) em Comunidades Vulneráveis
Considerando a necessidade de busca de soluções inovadoras para reduzir a insegurança alimentar em comunidades vulneráveis, este estudo avalia uma proposta de criação de uma árvore fotovoltaica integrada aos sistemas de aquaponia e hidroponia, indo ao encontro do 2º ODS (Objetivo de Desenvolvimento Sustentável), Fome Zero. Nesta abordagem, a árvore fotovoltaica alimenta o sistema aquapônico, onde os peixes fornecem nutrientes para as plantas, e a hidroponia coexiste, devolvendo água limpa aos peixes. Essa interação sinérgica e circular visa otimizar a produção de alimentos de forma sustentável. A metodologia inclui levantamento bibliográfico, análise documental e adaptações locais, focando na aldeia indígena Nova Esperança, afetada pelo denominado “desastre de Mariana”. Os resultados esperados da proposta destacam a contribuição dessa abordagem multifacetada para promover sistemas alimentares sustentáveis para populações vulneráveis. A integração de tecnologias renováveis e participação comunitária, podem ser fundamentais para enfrentar desafios como a insegurança alimentar e alcançar o ODS 2.
Multiplexing technique using photodetector arrays for quasi-distributed intensity variation optical fiber sensors system
We address the problem of simultaneously estimating the water-oil interface and total levels in three-phase separator tanks by applying three machine learning methods: Multilayer Perceptron (MLP), Kolmogorov-Arnold Networks (KAN), and Random Forest (RF). Data was collected from Fiber Bragg Grating-based optical sensors and processed to suppress outliers using the Local Outlier Factor algorithm. Hyperparameters for each model were optimized using Grid Search, and their performance was compared. The trained models were also exposed to scenarios with different levels of noise, and performance was evaluated. The results suggest that KAN performs effectively in predicting liquid levels, achieving a Root Mean Square Error of less than 3 mm and a Mean Absolute Percentage Error below 0.3% in scenarios without noise. Both MLP and KAN exhibit similar accuracy when the noise level is up to 1%. However, the MLP model outperforms KAN in higher noisy scenarios. On the other hand, the RF model shows the least effectiveness overall in noisy environments, though it does maintain a relatively stable maximum error across different noise levels. Therefore, we demonstrate that KAN has advantages in low-noise scenarios compared to conventional MLP and RF models. Conversely, MLP is more effective under higher noise conditions. These findings can aid in the research and development of monitoring systems.
Robustness of Liquid Neural Networks to Noise: A Comparative Study with Convolutional Neural Networks
This study presents a comparative evaluation of Convolutional Neural Networks (CNNs), Liquid Neural Networks (LNNs), and Multi-Layer Perceptrons (MLPs) for fluid level estimation using Fiber Bragg Grating (FBG) sensor data. Experiments were conducted with water and oil in a controlled tank environment, generating approximately 8,300 samples comprising amplitude and wavelength measurements from six sensors per fiber. Each model underwent hyperparameter optimization and was assessed under noise-free conditions as well as with Gaussian and 1/f colored noise at varying signal-to-noise ratios (10, 30, 50 dB). Results indicate that while all models achieved high accuracy on clean data, CNNs and MLPs were highly sensitive to strong noise. In contrast, LNNs demonstrated superior robustness and stability, maintaining low RMSE and high R2 across all noise levels. These findings highlight the LNN’s suitability for real-world fluid sensing applications where measurements are subject to environmental variability.