Lower Mean Research Articles

Integrating Machine Learning and the Finite Element Method for Assessing Stiffness Degradation in Photovoltaic Modules.

This study introduces a novel machine learning (ML) method utilizing a stacked auto-encoder network to predict stiffness degradation in photovoltaic (PV) modules with pre-existing cracks. The input data for the training process was derived from numerical simulations, ensuring a comprehensive representation of module behavior under various conditions. The findings highlight the robust predictive capability of the model, as evidenced by its impressive R2 value of 0.961 and notably low root mean square error (RMSE) of 4.02%. These metrics significantly outperform those of other conventional methods, including the convolutional neural network (CNN) with R2 of 0.905 and RMSE of 9.43%, the space vector machine (SVM) with R2 of 0.827 and RMSE of 17.93%, and the random forest (RF) with R2 of 0.899 and RMSE of 11.02%. Moreover, the findings suggest that the predictive dynamics of degradation are affected by the varying weight functions of different input parameters, such as climate temperature, grain size, material effort, and pre-crack size, as the degradation level changes. Furthermore, a geometric analysis reveals model deficiencies where significant overestimations correlate with thicker glass components, while pronounced underestimations are predominantly associated with thinner layers of polycrystalline silicon wafer and Ethylene Vinyl Acetate in the module. As a case study, it demonstrated that to maintain a constant degradation level between 1.30 and 1.32 in a PV module with components featuring consistent geometric attributes, the input parameters must be kept within specific ranges: climate temperature ranging from 33 to 57°C, grain size ranging from 36 to 81 μm, material effort ranging from 0.74 to 0.81, and pre-crack size ranging from 24 to 32 μm. Therefore, this underscores that the ML model not only predicts degradation but also delineates the parameter space required to achieve a consistent output value.

Coomans plot classification of lard in the ink blends and ink extracts from printed packaging films using lard Fourier transform infrared spectral profile and multivariate analysis

Lard in blends of commercial gravure ink (ranging from 0.5 to 20%) and ink extracts from eleven commercially printed packaging films for foodstuffs was characterized using Fourier transform infrared (FTIR) spectroscopy. The FTIR spectral bands at 4000 ‒ 650 cm-1 associated with the lard fingerprint were acquired and used to classify the lard and its blends using partial least squares (PLS) regression and discriminant analysis (DA). Commercial gravure inks (also used for preparing calibration curve samples), blends of lard ranging from 0.5 to 20% in gravure inks, and commercially printed food packaging films were tested. Linear correlation of predicted and actual values of lard were determined using PLS calibration and validation models. They produced a high coefficient of determination (R2 ) of 0.943, a low root mean square error of calibration (RMSEC) of 1.674, as well as a high R2 = 0.999 and a low root mean square error of prediction (RMSEP) of 1.233, respectively. The PLS calibration was verified employing a leave-oneout cross-validation, while DA was used to classify a series of lard standard, lard-added ink, and commercial food packaging films. The Coomans plot classification of the lardadded ink and commercial food packaging films illustrated that the food packaging samples were plotted in the right and left hemispheres in the lard-added ink class. This result also demonstrated that FTIR coupled with chemometric PLS predicted the lard content in the printing inks with high overall accuracy, as indicated by a low mean difference (MD) value of 0.577 and a low standard deviation of difference (SDD) value of 0.599. The DA allowed the ink packaging samples that potentially contained lard to be distinguished from those without lard. Sample 7 (commercially printed food packaging ink) exhibits the highest possibility of containing lard.

Pareto optimization of SPECT acquisition and reconstruction settings for 177Lu activity quantification

BackgroundThe aim was to investigate the noise and bias properties of quantitative 177Lu-SPECT with respect to the number of projection angles, and the number of subsets and iterations in the OS-EM reconstruction, for different total acquisition times.MethodsExperimental SPECT acquisition of six spheres in a NEMA body phantom filled with 177Lu was performed, using medium-energy collimators and 120 projections with 180 s per projection. Bootstrapping was applied to generate data sets representing acquisitions with 20 to 120 projections for 10 min, 20 min, and 40 min, with 32 noise realizations per setting. Monte Carlo simulations were performed of 177Lu-DOTA-TATE in an anthropomorphic computer phantom with three tumours (2.8 mL to 40.0 mL). Projections representing 24 h and 168 h post administration were simulated, each with 32 noise realizations. Images were reconstructed using OS-EM with compensation for attenuation, scatter, and distance-dependent resolution. The number of subsets and iterations were varied within a constrained range of the product number of iterations ×\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\ imes$$\\end{document} number of projections ≤2400\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\le 2400$$\\end{document}. Volumes-of-interest were defined following the physical size of the spheres and tumours, the mean activity-concentrations estimated, and the absolute mean relative error and coefficient of variation (CV) over noise realizations calculated. Pareto fronts were established by analysis of CV versus mean relative error.ResultsPoints at the Pareto fronts with low CV and high mean error resulted from using a low number of subsets, whilst points at the Pareto fronts associated with high CV but low mean error resulted from reconstructions with a high number of subsets. The number of projection angles had limited impact.ConclusionsFor accurate estimation of the 177Lu activity-concentration from SPECT images, the number of projection angles has limited importance, whilst the total acquisition time and the number of subsets and iterations are parameters of importance.

Development and validation of the Thai version of MNREAD acuity charts.

To develop and validate MNREAD acuity charts in the Thai language (MNREAD-TH). In total, 180 Thai sentences were generated using words specific to the primary 1-3 students. Each sentence contained 60 characters divided into three lines with even left and right margins and was evaluated by instructors in the education program in Thai. The selected 118 sentences were tested in 20 adults and 20 children by measuring reading time and recording the errors. Sentences with extremely high or low mean reading speeds were excluded. The remaining sentences were selected for contributing to the MNREAD-TH charts. For validation, the charts were tested in another 20 normal-sighted adult groups. In total, 118 sentences were tested on 20 adults (range: 23-58 years) and 20 primary school students (age: 8-9 years). The mean (SD) reading speeds in adult and children's groups were 134.09 (12.45) and 71.33 (10.77) words/min (wpm), respectively. After removing the sentences that deviated by ±1 SD from the mean reading speed, repeatedly had difficulty words, and had low subjective scoring, 60 matched sentences between the groups were selected to develop three versions of the MNREAD-TH chart. The intraclass correlation coefficients between charts were 0.85 for reading acuity, 0.94 for maximum reading speed, and 0.79 for critical print size. The MNREAD-TH charts can be reliably used for evaluating reading performance in Thai people for both children and adults with either normal sight or low vision. These charts are standardized and helpful in clinical trials involving reading.

Prediction and interpretation of limit pressure of clayey soils using ensemble machine learning methods and shapely additive explanations

The pressuremeter test (PMT), a valuable geotechnical in situ test, is used to design foundations of varying depths (shallow, semi-deep, and deep). It assesses a soil's bearing capacity and settlement through two key parameters: limit pressure and pressuremeter modulus. However, the high cost and time demands of PMTs limit their widespread use. This study addresses this challenge by exploring the effectiveness of ensemble machine learning algorithms. To achieve this main goal, we employ two methods, Extreme Gradient Boosting (XGBoost) and Random Forest, to predict limit pressure of soil. To develop the mentioned models an experimental database was used to train and validate the developed models. The effectiveness of these methods are evaluated using three statistical metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Coefficient of Determination (R²). The performance metrics show that the developed XGBoost and Random Forest models are viable alternatives to the pressure meter test for estimating limit pressure. Both models achieved high R-squared values (around 0.99 for training and 0.90 for testing) and a low root mean squared error (RMSE) of 3.23 and 4.13 for the testing set, respectively. These results demonstrate the effectiveness of using machine learning in the geotechnical field. To further understand the influence of individual factors on the predictions, we will utilize the Shapley Additive explanations (SHAP) method. This technique analyzes the contribution of each input variable (feature) to the model's predictions of limit pressure. By quantifying the importance of these features; SHAP provides valuable insights into which soil properties most significantly affect the foundation design parameters.

Neural Networks For Particle Diffusometry In The Presence Of Flow And With Defocused Particles

Diffusion is a natural phenomenon in fluids. Its measurement can be done optically by seeding an otherwise featureless fluid with tracer particles and observing their motion. However, existing algorithms for particle-based diffusion coefficient measurement have multiple failure modes especially when the fluid has a flow or the particles are defocused. We present a method based on Convolutional Neural Networks (CNNs) for predicting diffusion coefficients in the presence of these real-world effects. The CNNs were trained, validated, and tested on crops from four-frame temporally averaged images. The study includes three simulated datasets: Gaussian-shaped particles under no fluid flow condition, Gaussian-shaped particles under an arbitrary flow condition, and defocused particles under no fluid flow condition. The results show that the CNNs have a low Mean Absolute Error (MAE) of 0.12 µm 2 /s between the true and predicted diffusion coefficient values for the dataset with Gaussian-shaped particles under no fluid flow condition. The CNNs have a slightly higher MAE of 0.19 µm^2 /s for Gaussian-shaped particles under arbitrary flow and defocused particles under no fluid flow conditions. The performance of the CNNs was also benchmarked against four conventional algorithms on these simulated datasets. The results show that the conventional algorithms perform better than the CNNs when the particles are assumed to be Gaussian and the fluid has no flow. However, the CNNs outperform conventional methods in the presence of flow or if the particles are defocused. Finally, the outputs of CNNs were compared against the outputs of conventional algorithms on experimental datasets. This provides uncertainty in the range of 0.19 µm^2 /s - 0.47 µm^2 /s, which is slightly larger than the errors obtained from simulated datasets. Hence, the study shows that CNNs can be used to reliably predict diffusion coefficients from complex particle datasets with as little as four frames.

A sensorless centrifugal pump operation states estimation method based on HSSA-BPNN

As a liquid transportation equipment, centrifugal pumps find extensive use in industrial process control. Certain environments involving high temperature, high pressure, and highly corrosive media may not permit the installation of flow and pressure sensors. Consequently, accurately estimating the operating status of centrifugal pumps in the absence of sensors has become a prominent research focus. However, the precision of existing sensorless estimation methods remains a concern. This study introduces a HSSA-BPNN model utilizing valve opening degree, motor rotational speed and shaft power as input variables. A hybrid strategy improved sparrow search algorithm (HSSA) is developed to optimize the back propagation neural network (BPNN). Experimental results demonstrate a low mean absolute percentage error (MAPE) of the test data, which are 2.2741% and 1.6502%, respectively. The absolute errors of flow rate and head are below 0.3 m3/h and 0.5 m, respectively, with relative errors less than 4% and 3%, meeting practical applications.

Extending the historical in situ gauge network over Indian river reaches utilizing the Surface Water and Ocean Topography (SWOT) mission

ABSTRACT We present a preliminary study on using the Surface Water and Ocean Topography (SWOT) mission to extend water surface elevation (WSE) measurements from a sparsely distributed in situ gauge network along a one-dimensional (1D) study reach within the Mahanadi River basin in India. We use the Centre National d’Etudes Spatiales (CNES) simulator to simulate 3 years of synthetic measurements of SWOT WSE, which are assimilated into a 1D stochastic model based on six independent configurations. For each configuration, the model is derived using historical measurements of in situ WSE to simulate daily WSE estimates every 200 m along the study reach. Assimilated outputs showed high Nash-Sutcliffe efficiency (0.697–0.811), high correlation (0.838–0.901) and low root mean square error (0.242–0.513 m) with respect to in situ validation data. Results indicate consistent performance of the SWOT-assimilated model framework and are relevant considering the declining availability of in situ monitoring stations.

Extending the historical in situ gauge network over Indian river reaches utilizing the Surface Water and Ocean Topography (SWOT) mission

ABSTRACT We present a preliminary study on using the Surface Water and Ocean Topography (SWOT) mission to extend water surface elevation (WSE) measurements from a sparsely distributed in situ gauge network along a one-dimensional (1D) study reach within the Mahanadi River basin in India. We use the Centre National d’Etudes Spatiales (CNES) simulator to simulate 3 years of synthetic measurements of SWOT WSE, which are assimilated into a 1D stochastic model based on six independent configurations. For each configuration, the model is derived using historical measurements of in situ WSE to simulate daily WSE estimates every 200 m along the study reach. Assimilated outputs showed high Nash-Sutcliffe efficiency (0.697–0.811), high correlation (0.838–0.901) and low root mean square error (0.242–0.513 m) with respect to in situ validation data. Results indicate consistent performance of the SWOT-assimilated model framework and are relevant considering the declining availability of in situ monitoring stations.

In the Qaidam Basin, Soil Nutrients Directly or Indirectly Affect Desert Ecosystem Stability under Drought Stress through Plant Nutrients.

The low nutrient content of soil in desert ecosystems results in unique physiological and ecological characteristics of plants under long-term water and nutrient stress, which is the basis for the productivity and stability maintenance of the desert ecosystem. However, the relationship between the soil and the plant nutrient elements in the desert ecosystem and its mechanism for maintaining ecosystem stability is still unclear. In this study, 35 sampling sites were established in an area with typical desert vegetation in the Qaidam Basin, based on a drought gradient. A total of 90 soil samples and 100 plant samples were collected, and the soil's physico-chemical properties, as well as the nutrient elements in the plant leaves, were measured. Regression analysis, redundancy analysis (RDA), the Theil-Sen Median and Mann-Kendall methods, the structural equation model (SEM), and other methods were employed to analyze the distribution characteristics of the soil and plant nutrient elements along the drought gradient and the relationship between the soil and leaf nutrient elements and its impact on ecosystem stability. The results provided the following conclusions: Compared with the nutrient elements in plant leaves, the soil's nutrient elements had a more obvious regularity of distribution along the drought gradient. A strong correlation was observed between the soil and leaf nutrient elements, with soil organic carbon and alkali-hydrolyzed nitrogen identified as important factors influencing the leaf nutrient content. The SEM showed that the soil's organic carbon had a positive effect on ecosystem stability by influencing the leaf carbon, while the soil's available phosphorus and the mean annual temperature had a direct positive effect on stability, and the soil's total nitrogen had a negative effect on stability. In general, the soil nutrient content was high in areas with a low mean annual temperature and high precipitation, and the ecosystem stability in the area distribution of typical desert vegetation in the Qaidam Basin was low. These findings reveal that soil nutrients affect the stability of desert ecosystems directly or indirectly through plant nutrients in the Qaidam Basin, which is crucial for maintaining the stability of desert ecosystems with the background of climate change.

Learning carotid vessel wall segmentation in black-blood MRI using sparsely sampled cross-sections from 3D data.

Atherosclerosis of the carotid artery is a major risk factor for stroke. Quantitative assessment of the carotid vessel wall can be based on cross-sections of three-dimensional (3D) black-blood magnetic resonance imaging (MRI). To increase reproducibility, a reliable automatic segmentation in these cross-sections is essential. We propose an automatic segmentation of the carotid artery in cross-sections perpendicular to the centerline to make the segmentation invariant to the image plane orientation and allow a correct assessment of the vessel wall thickness (VWT). We trained a residual U-Net on eight sparsely sampled cross-sections per carotid artery and evaluated if the model can segment areas that are not represented in the training data. We used 218 MRI datasets of 121 subjects that show hypertension and plaque in the ICA or CCA measuring in ultrasound. The model achieves a high mean Dice coefficient of 0.948/0.859 for the vessel's lumen/wall, a low mean Hausdorff distance of , and a low mean average contour distance of on the test set. The model reaches similar results for regions of the carotid artery that are not incorporated in the training set and on MRI of young, healthy subjects. The model also achieves a low median Hausdorff distance of on the 2021 Carotid Artery Vessel Wall Segmentation Challenge test set. The proposed method can reduce the effort for carotid artery vessel wall assessment. Together with human supervision, it can be used for clinical applications, as it allows a reliable measurement of the VWT for different patient demographics and MRI acquisition settings.

Machine learning for parameters diagnosis of spark discharge by electro-acoustic signal

Discharge plasma parameter measurement is a key focus in low-temperature plasma research. Traditional diagnostics often require costly equipment, whereas electro-acoustic signals provide a rich, non-invasive, and less complex source of discharge information. This study harnesses machine learning to decode these signals. It establishes links between electro-acoustic signals and gas discharge parameters, such as power and distance, thus streamlining the prediction process. By building a spark discharge platform to collect electro-acoustic signals and implementing a series of acoustic signal processing techniques, the Mel-Frequency Cepstral Coefficients (MFCCs) of the acoustic signals are extracted to construct the predictors. Three machine learning models (Linear Regression, k-Nearest Neighbors, and Random Forest) are introduced and applied to the predictors to achieve real-time rapid diagnostic measurement of typical spark discharge power and discharge distance. All models display impressive performance in prediction precision and fitting abilities. Among them, the k-Nearest Neighbors model shows the best performance on discharge power prediction with the lowest mean square error (MSE = 0.00571) and the highest -squared value ( ). The experimental results show that the relationship between the electro-acoustic signal and the gas discharge power and distance can be effectively constructed based on the machine learning algorithm, which provides a new idea and basis for the online monitoring and real-time diagnosis of plasma parameters.

An improved meteorological variables-based aerosol optical depth estimation method by combining a physical mechanism model with a two-stage model

A two-stage model integrating a spatiotemporal linear mixed effect (STLME) and a geographic weight regression (GWR) model is proposed to improve the meteorological variables-based aerosol optical depth (AOD) retrieval method (Elterman retrieval model-ERM). The proposed model is referred to as the STG-ERM model. The STG-ERM model is applied over the Beijing-Tianjin-Hebei (BTH) region in China for the years 2019 and 2020. The results show that data coverage increased by 39.0% in 2019 and 40.5% in 2020. Cross-validation of the retrieval results versus multi-angle implementation of atmospheric correction (MAIAC) AOD shows the substantial improvement of the STG-ERM model over earlier meteorological models for AOD estimation, with a determination coefficient (R2) of daily AOD of 0.86, root mean squared prediction error (RMSE) and the relative prediction error (RPE) of 0.10 and 36.14% in 2019 and R2 of 0.86, RMSE of 0.12 and RPE of 37.86% in 2020. The fused annual mean AOD indicates strong spatial variation with high value in south plain and low value in northwestern mountainous areas of the BTH region. The overall spatial seasonal mean AOD ranges from 0.441 to 0.586, demonstrating strongly seasonal variation. The coverage of STG-ERM retrieved AOD, as determined in this exercise by leaving out part of the meteorological data, affects the accuracy of fused AOD. The coverage of the meteorological data has smaller impact on the fused AOD in the districts with low annual mean AOD of less than 0.35 than that in the districts with high annual mean AOD of greater than 0.6. If available, continuous daily meteorological data with high spatiotemporal resolution can improve the model performance and the accuracy of fused AOD. The STG-ERM model may serve as a valuable approach to provide data to fill gaps in satellite-retrieved AOD products.

Microstructure Analysis, Piezoelectrical Resistivity, and Compressive Strength Concrete Incorporated with Waste Steel Slag as a Fine Aggregate Replacement

Abstract This study aims to investigate the effect of waste steel slag (SS) as partially replaced with cement and fine aggregate on conventional concrete for different mixes named M25, M35, and M47 in terms of compressive strength (CS), electrical resistivity (ER), and piezoresistivity behavior. SS is a molten mixture of silicates and oxides that solidifies upon cooling, a byproduct of the steel-making process. Before doing the design experiments, the optimum value of SS as powder and fine aggregate was determined using seven different mixes to investigate the effect of different SS sizes on the CS and piezoresistivity of normal concrete. Based on the results achieved, the optimum value and size of SS were selected to modify and investigate the effect of SS on three different mixes of conventional concrete named M25, M35, and M47 in terms of CS, ER, and piezoresistivity behavior. The resistivity of all concrete mixes was measured using four-probe from early curing to 28 days of curing time. The results demonstrated that M47 mix modified with SS has lower resistivity than the rest of the concrete mixes. The results of piezoresistivity behavior indicated that M47 mix modified with SS has a higher resistivity change while applying stress at 3 days of curing compared to the M25 and M35 concrete mix modified with SS by 44.1 % and 37.6 %, respectively. The Vipulandanan p-q model was applied to predict both ER versus time and change of resistivity versus stress for all mixes. The results demonstrated that the model predicted the change of resistivity versus applied stress with a high coefficient of determination that varied between 0.82 and 0.989, and a low root mean square error changed between 0.81 Ω.m and 7.94 Ω.m.

Enhancing residential heating load prediction with advanced machine learning and optimization techniques

The effective prediction of heating load in residential buildings is crucial for optimizing energy consumption and enhancing thermal comfort. This study investigates three prominent regression techniques Extra Tree Regression, Decision Tree Regression, and Support Vector Regression utilized for this purpose. To improve their predictive performance, a novel metaheuristic optimization algorithm, the Tyrannosaurus Rex Optimization Algorithm, is employed. The dataset used includes features such as Relative Compactness, Surface Area, Wall Area, Roof Area, Overall Height, Orientation, Glazing Area, and Glazing Area Distribution. Each machine learning model (Extra Tree Regression, Decision Tree Regression, Support Vector Regression) is trained on historical heating load data and optimized using the Tyrannosaurus Rex Optimization Algorithm to adjust internal parameters. Comparative analysis demonstrates that the proposed Extra Tree + Tyrannosaurus Rex Optimization Algorithm model achieves superior performance with an R-squared value of 0.987 and a low Root Mean Square Error of 1.167. By leveraging the strengths of different machine learning algorithms, this research presents a reliable approach for accurately predicting heating load in residential buildings. This method holds promise for improving energy efficiency, enhancing thermal comfort, and promoting sustainable building design.

An improved meteorological variables-based aerosol optical depth estimation method by combining a physical mechanism model with a two-stage model

A two-stage model integrating a spatiotemporal linear mixed effect (STLME) and a geographic weight regression (GWR) model is proposed to improve the meteorological variables-based aerosol optical depth (AOD) retrieval method (Elterman retrieval model-ERM). The proposed model is referred to as the STG-ERM model. The STG-ERM model is applied over the Beijing-Tianjin-Hebei (BTH) region in China for the years 2019 and 2020. The results show that data coverage increased by 39.0% in 2019 and 40.5% in 2020. Cross-validation of the retrieval results versus multi-angle implementation of atmospheric correction (MAIAC) AOD shows the substantial improvement of the STG-ERM model over earlier meteorological models for AOD estimation, with a determination coefficient (R2) of daily AOD of 0.86, root mean squared prediction error (RMSE) and the relative prediction error (RPE) of 0.10 and 36.14% in 2019 and R2 of 0.86, RMSE of 0.12 and RPE of 37.86% in 2020. The fused annual mean AOD indicates strong spatial variation with high value in south plain and low value in northwestern mountainous areas of the BTH region. The overall spatial seasonal mean AOD ranges from 0.441 to 0.586, demonstrating strongly seasonal variation. The coverage of STG-ERM retrieved AOD, as determined in this exercise by leaving out part of the meteorological data, affects the accuracy of fused AOD. The coverage of the meteorological data has smaller impact on the fused AOD in the districts with low annual mean AOD of less than 0.35 than that in the districts with high annual mean AOD of greater than 0.6. If available, continuous daily meteorological data with high spatiotemporal resolution can improve the model performance and the accuracy of fused AOD. The STG-ERM model may serve as a valuable approach to provide data to fill gaps in satellite-retrieved AOD products.

A sensorless centrifugal pump operation states estimation method based on HSSA-BPNN

As a liquid transportation equipment, centrifugal pumps find extensive use in industrial process control. Certain environments involving high temperature, high pressure, and highly corrosive media may not permit the installation of flow and pressure sensors. Consequently, accurately estimating the operating status of centrifugal pumps in the absence of sensors has become a prominent research focus. However, the precision of existing sensorless estimation methods remains a concern. This study introduces a HSSA-BPNN model utilizing valve opening degree, motor rotational speed and shaft power as input variables. A hybrid strategy improved sparrow search algorithm (HSSA) is developed to optimize the back propagation neural network (BPNN). Experimental results demonstrate a low mean absolute percentage error (MAPE) of the test data, which are 2.2741% and 1.6502%, respectively. The absolute errors of flow rate and head are below 0.3 m3/h and 0.5 m, respectively, with relative errors less than 4% and 3%, meeting practical applications.

Meander Structure Analysis Techniques Using Artificial Neural Networks

Typically, analyses of meander structures (MSs) for transfer characteristics are conducted using specialized commercial software based on numerical methods. However, these methods can be time-consuming, particularly when a researcher is seeking to perform a preliminary study of the designed structures. This study aims to explore the application of neural networks in the design and analysis of meander structures. Three different feedforward neural network (FFNN), time delay neural network (TDNN), and convolutional neural network (CNN) techniques were investigated for the analysis and design of the meander structures in this article. The geometric dimensions or top-view images of 369 different meander structures were used for training an FFNN, TDNN, and CNN. The investigated networks were designed to predict such electrodynamic parameters as the delay time (td), reflection coefficient (S11), and transmission coefficient (S21) in the 0–10 GHz frequency band. A sufficiently low mean absolute error (MAE) was achieved with all three methods for the analysis of MSs. Using an FFNN, the characteristic td was predicted with a 3.3 ps average MAE. The characteristic S21 was predicted with a 0.64 dB average MAE, and S11 was predicted with a 2.47 dB average MAE. The TDNN allowed the average MAEs to be reduced to 0.9 ps, 0.11 dB, and 1.63 dB, respectively. Using a CNN, the average MAEs were 27.5 ps, 0.44 dB, and 1.36 dB, respectively. The use of neural networks has allowed accelerating the analysis procedure from approximately 120 min on average to less than 5 min.

The Differences on the Fatty Infiltration of Paraspinal Muscles between Single- and Multiple-level Intervertebral Disc Degeneration in Patients with Lumbar Disc Herniation.

Multiple-level Intervertebral disc degeneration (IDD) in patients with lumbar disc herniation (LDH) is related to postoperative re-herniation and low back pain. Although many investigators believed that there is an interdependence between paraspinal muscles degeneration and IDD, few studies focused on the fatty infiltration of paraspinal muscles on single- and multiple-level IDD in patients with LDH. This study aims to investigate the difference on the fatty infiltration of paraspinal muscles between single- and multiple-levels IDD in patients with LDH. and to explore in patients with LDH whether fatty infiltration is a potential risk factor for multiple-level IDD. This study was conducted as a retrospective observational analysis of 82 patients with LDH from January 1, 2020 to December 30, 2020 in our hospital were enrolled. Twenty-seven cases had single-level IDD (Group A), and 55 cases had multiple-level IDD (Group B). We measured the mean computed tomography (CT) density value of the paraspinal muscles, including multifidus (MF), erector spinae (ES) and psoas muscle (PM) at each disc from L1 to S1. Subgroups were set to further analyze the odds ratio (OR) of fatty infiltration of paraspinal muscles in different sex and BMI groups. We measured sagittal angles and analyzed the relationships between these angles and IDD. Finally, we use logistic regression, adjusted for other confounding factors, to investigate whether fatty infiltration is an independent risk factor for multi-level IDD. The average age in multi-level IDD (51.40 ± 15.47 years)was significantly higher than single-level IDD (33.37 ± 7.10 years). The mean CT density value of MF, ES and PM in single-level IDD was significantly higher than multi-level IDD (all ps < 0.001). There was no significant difference of the mean value of angles between the two groups. No matter being fat (body mass index [BMI] > 24.0 kg/m2) or normal, patients with low mean muscle CT density value of MF and ES are significantly easier to suffer from multiple-level IDD. In the pure model, the average CT density value of the MF, ES and PM is all significantly associated with the occurrence of multi-IDD. However, after adjusting for various confounding factors, only the OR of the average CT density value for MF and ES remains statistically significant (OR = 0.810, 0.834, respectively). In patients with LDH, patients with multiple-level IDD have more severe fatty infiltration of MF and ES than those with single-level IDD. Fatty infiltration of MF and ES are independent risk factors for multiple-level IDD in LDH patients.

Determination and Prediction of Corrosion Integrity on RV Discovery Via Non-Destructive Test (NDT) And Artificial Neural Network (ANN)

Marine environment is a harsh and severe environment for a metal structure like vessel, oil rigs and port infrastructure. A regular survey or monitoring is needed to reduce a structure failure due to the corrosion. To be seaworthy, a vessel should undergo a regular survey under specified timeframe. This survey is time consuming and costly. An alternative approach is required to predict the structural integrity of a vessel. Artificial Neural Network is one of the current methods that can be used to predict the deterioration rate of a structure. Corrosion integrity of RV-Discovery was determined via plate thickness measurement, coating thickness measurement and potential measurement. The data obtained from these measurements were used in artificial neural network to predict the deterioration rate. The results indicate that the plate and coating thickness reduction percentage is within minimal range while the average potential changes show that the structure is in passivation state. It implies that the structural integrity is in a good state with no or minimal maintenance required. The prediction of deterioration rate also shows that Scaled Conjugate Gradient (SCG) training algorithm was able to predict with over 95% of confidence and low mean square error.