Data-driven Machine Learning Approach Research Articles

Predicting the Curie temperature of Sm-Co-based alloys via data-driven strategy

Calculating the Curie temperature of rare-earth permanent magnetic materials has remained a big theoretical challenge. In this study, based on a home-built Sm-Co-based alloys database, a data-driven machine learning approach was developed to predict the Curie temperature of Sm-Co-based alloys. High-throughput predictions of Curie temperature were achieved using a genetic program based symbolic regression model. A classification model based on logistic regression was established to quantify the effect of doping on the Curie temperature of Sm-Co-based alloys. The key physical descriptor affecting Curie temperature was extracted from the established machine learning models, and the Curie temperature sensitivity coefficient was defined. It was discovered that the doping elements with large electrical conductivity and similar heat of fusion to that of Sm are likely to increase the Curie temperature of Sm-Co-based alloys. The model predictions were verified quantitatively by the experimental results of a series of prepared Sm-Co-based samples. This work provides a high-efficiency method for developing Sm-Co-based permanent magnets with high Curie temperatures.

NOx Emission Predictions in Gas Turbines Through Integrated Data-Driven Machine Learning Approaches

Abstract The reduction of NOx emissions is a paramount endeavor in contemporary engineering and energy production, as these emissions are closely linked to adverse environmental and health impacts. The prediction of NOx emission from gas turbines through several integrated data-driven machine learning methods has been evaluated in study. The study compares the performance of ensemble and conventional machine learning models, demonstrating superior accuracy achieved by the ensemble models. Specifically, the Random Forest model achieved an accuracy rate of 91.68%, XGBoost yielded an accuracy of 91.54%, and CATBoost exhibited the highest accuracy at 92.76%. These findings highlight the capability of data-driven machine learning techniques in enhancing NOx emission predictions in gas turbines. The improved prediction by ensembles can be utilized in the development and implementation of more effective control and mitigation strategies in practical applications. Through the application of these advanced machine learning approaches, the gas turbine industry can play a pivotal role in minimizing its environmental impact while optimizing operational efficiency. This study also provides valuable insights into the effectiveness of ensemble machine learning models, advancing our understanding of their capabilities in addressing the critical issue of NOx emissions from gas turbines.

Amyloid-β prediction machine learning model using source-based morphometry across neurocognitive disorders

Previous studies have developed and explored magnetic resonance imaging (MRI)-based machine learning models for predicting Alzheimer’s disease (AD). However, limited research has focused on models incorporating diverse patient populations. This study aimed to build a clinically useful prediction model for amyloid-beta (Aβ) deposition using source-based morphometry, using a data-driven algorithm based on independent component analyses. Additionally, we assessed how the predictive accuracies varied with the feature combinations. Data from 118 participants clinically diagnosed with various conditions such as AD, mild cognitive impairment, frontotemporal lobar degeneration, corticobasal syndrome, progressive supranuclear palsy, and psychiatric disorders, as well as healthy controls were used for the development of the model. We used structural MR images, cognitive test results, and apolipoprotein E status for feature selection. Three-dimensional T1-weighted images were preprocessed into voxel-based gray matter images and then subjected to source-based morphometry. We used a support vector machine as a classifier. We applied SHapley Additive exPlanations, a game-theoretical approach, to ensure model accountability. The final model that was based on MR-images, cognitive test results, and apolipoprotein E status yielded 89.8% accuracy and a receiver operating characteristic curve of 0.888. The model based on MR-images alone showed 84.7% accuracy. Aβ-positivity was correctly detected in non-AD patients. One of the seven independent components derived from source-based morphometry was considered to represent an AD-related gray matter volume pattern and showed the strongest impact on the model output. Aβ-positivity across neurological and psychiatric disorders was predicted with moderate-to-high accuracy and was associated with a probable AD-related gray matter volume pattern. An MRI-based data-driven machine learning approach can be beneficial as a diagnostic aid.

Data-Driven Machine Learning Approach for Modeling the Production and Predicting the Characteristics of Aligned Electrospun Nanofibers

Data-Driven Machine Learning Approach for Modeling the Production and Predicting the Characteristics of Aligned Electrospun Nanofibers

Next-generation non-linear and collapse prediction models for short- to long-period systems via machine learning methods

Since the 1960 s, a cornerstone of earthquake engineering has been estimating the non-linear response of structures based just on lateral strength, modal properties, and the anticipated seismic demand. Over the years, several studies have quantified this empirical relationship and integrated it within seismic design and assessment methodologies. These have been widely accepted for practical application and adopted in building codes worldwide. While these models work reasonably well, there are still areas in which improvements can be made, especially concerning their robust quantification of uncertainty. This is mainly due to the amount of data used to quantify these empirical relationships, the choice of functional forms during the fitting, the model fitting and testing process, and how the ground motion shaking intensity is characterised. This study tackles these issues via the non-linear analysis of single-degree-of-freedom oscillators to train several machine learning (ML) models. This was to examine the accuracy and applicability of such models within a seismic engineering context and explore potential gains in quantifying the non-linear response of structures via next-generation intensity measures, namely average spectral acceleration, Saavg. The results show that the Decision Tree and XGBoost models worked well across a broad range of periods, accurately predicting collapse and non-collapse responses. Appraising these with existing models showed a notable improvement all around. It indicates that the models based on data-driven ML approaches represent a positive step and can be seamlessly integrated with seismic analysis methodologies utilised worldwide. Subsequently, a Python-based library of these XGBoost ML models was developed and made available online to foster practical application.

Knee-Point-Conscious Battery Aging Trajectory Prediction Based on Physics-Guided Machine Learning

Early prediction of aging trajectories of lithium-ion (Li-ion) batteries is critical for cycle life testing, quality control, and battery health management. Although data-driven machine learning (ML) approaches are well suited for this task, unfortunately, relying solely on data is exceedingly time-consuming and resource-intensive, even in accelerated aging with complex aging mechanisms. This challenge is rooted in the highly complex and time-varying degradation mechanisms of Li-ion battery cells. We propose a novel method based on physics-guided machine learning (PGML) to overcome this issue. First, electrode-level physical information is incorporated into the model training process to predict the aging trajectory’s knee point (KP). The relationship between the identified KP and the accelerated aging behavior is then explored, and an aging trajectory prediction algorithm is developed. The prior knowledge of aging mechanisms enables a transfer of valuable physical insights to yield accurate KP predictions with small data and weak correlation feature relationship. Based on a Li[NiCoMn]O <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> cell dataset, we demonstrate that only 14 cells are needed to train a PGML model for achieving a lifetime prediction error of 2.02% using the data of the first 50 cycles. In contrast, at least 100 cells are needed to reach this level of accuracy without the physical insights.

Predicting ocean pressure field with a physics-informed neural network.

Ocean sound pressure field prediction, based on partially measured pressure magnitudes at different range-depths, is presented. Our proposed machine learning strategy employs a trained neural network with range-depth as input and outputs complex acoustic pressure at the location. We utilize a physics-informed neural network (PINN), fitting sampled data while considering the additional information provided by the partial differential equation (PDE) governing the ocean sound pressure field. In vast ocean environments with kilometer-scale ranges, pressure fields exhibit rapidly fluctuating phases, even at frequencies below 100 Hz, posing a challenge for neural networks to converge to accurate solutions. To address this, we utilize the envelope function from the parabolic-equation technique, fundamental in ocean sound propagation modeling. The envelope function shows slower variations across ranges, enabling PINNs to predict sound pressure in an ocean waveguide more effectively. Additional PDE information allows PINNs to capture PDE solutions even with a limited amount of training data, distinguishing them from purely data-driven machine learning approaches that require extensive datasets. Our approach is validated through simulations and using data from the SWellEx-96 experiment.

Unsupervised deep learning approach for structural anomaly detection using probabilistic features

Civil structures may deteriorate during their service life due to degradation or damage imposed by natural hazards such as earthquakes, wind, and impact. Structural performance anomaly detection is essential to provide an early warning of structural degradation limit states in order to prevent potential catastrophic failure. Data-driven machine learning approaches have been widely used for this, due to their capability in capturing features sensitive to damage-induced anomalies from structural health monitoring (SHM) data, assuming that such data are available. Although machine learning models have been used, many are challenged by the vast operational and environmental variability that can corrupt SHM data and by (typically) strongly correlated information from different sensors in the SHM data. This paper proposes an unsupervised deep learning approach for the detection of structural anomaly based on a deep convolutional variational autoencoder (DCVAE) for feature extraction coupled with support vector data description (SVDD) for anomaly detection. The proposed DCVAE-SVDD method has several appealing strengths. First, the variational latent encoding is used to capture the features of monitoring data through a probability distribution. The integration of the Kullback–Leibler divergence in the loss function provides accurate estimation of the probability distributions. Second, the DCVAE designed with convolutional and deconvolutional operations utilizes the correlation among multisensor data to avoid loss of correlation features and achieve better performance in feature extraction. Third, the SVDD is utilized to create a minimum-volume hypersphere that contains the anomaly-sensitive statistical features of the state. The hypersphere accurately separates anomaly-sensitive statistical features of reference states of structure from the anomalous ones. A computational frame model and a laboratory grandstand model are used to evaluate the performance of the proposed method for detecting structural anomaly. The results demonstrate the superiority of the proposed DCVAE-SVDD in detection accuracy over the other commonly used structural anomaly detection methods (deep autoencoder combined with SVDD autoregressive model with one-class support vector machine, and principal component analysis).

Machine Learning and Signal Processing for Bridge Traffic Classification with Radar Displacement Time-Series Data

This paper introduces a novel nothing-on-road (NOR) bridge weigh-in-motion (BWIM) approach with deep learning (DL) and non-invasive ground-based radar (GBR) time-series data. BWIMs allow site-specific structural health monitoring (SHM) but are usually difficult to attach and maintain. GBR measures the bridge deflection contactless. In this study, GBR and an unmanned aerial vehicle (UAV) monitor a two-span bridge in Germany to gather ground-truth data. Based on the UAV data, we determine vehicle type, lane, locus, speed, axle count, and axle spacing for single-presence vehicle crossings. Since displacement is a global response, using peak detection like conventional strain-based BWIMs is challenging. Therefore, we investigate data-driven machine learning approaches to extract the vehicle configurations directly from the displacement data. Despite a small and imbalanced real-world dataset, the proposed approaches classify, e.g., the axle count for trucks with a balanced accuracy of 76.7% satisfyingly. Additionally, we demonstrate that, for the selected bridge, high-frequency vibrations can coincide with axles crossing the junction between the street and the bridge. We evaluate whether filtering approaches via bandpass filtering or wavelet transform can be exploited for axle count and axle spacing identification. Overall, we can show that GBR is a serious contender for BWIM systems.

Machine Learning Prediction of Molecular Binding Profiles on Metal-Porphyrin via Spectroscopic Descriptors.

The study of molecular adsorption is crucial for understanding various chemical processes. Spectroscopy offers a convenient and non-invasive way of probing structures of adsorbed states and can be used for real-time observation of molecular binding profiles, including both structural and energetic information. However, deciphering atomic structures from spectral information using the first-principles approach is computationally expensive and time-consuming because of the sophistication of recording spectra, chemical structures, and their relationship. Here, we demonstrate the feasibility of a data-driven machine learning approach for predicting binding energy and structural information directly from vibrational spectra of the adsorbate by using CO adsorption on iron porphyrin as an example. Our trained machine learning model is not only interpretable but also readily transferred to similar metal-nitrogen-carbon systems with comparable accuracy. This work shows the potential of using structure-encoded spectroscopic descriptors in machine learning models for the study of adsorbed states of molecules on transition metal complexes.

Data-driven machine learning approaches for predicting permeability and corrosion risk in hybrid concrete incorporating blast furnace slag and fly ash

Data-driven machine learning approaches for predicting permeability and corrosion risk in hybrid concrete incorporating blast furnace slag and fly ash

A data-driven machine learning approach for electron-molecule ionization cross sections

Despite their importance in a wide variety of applications, the estimation of ionization cross sections for large molecules continues to present challenges for both experiment and theory. Machine learning (ML) algorithms have been shown to be an effective mechanism for estimating cross section data for atomic targets and a select number of molecular targets. We present an efficient ML model for predicting ionization cross sections for a broad array of molecular targets. Our model is a 3-layer neural network that is trained using published experimental datasets. There is minimal input to the network, making it widely applicable. We show that with training on as few as 10 molecular datasets, the network is able to predict the experimental cross sections of additional molecules with an accuracy similar to experimental uncertainties in existing data. As the number of training molecular datasets increased, the network’s predictions became more accurate and, in the worst case, were within 30% of accepted experimental values. In many cases, predictions were within 10% of accepted values. Using a network trained on datasets for 25 different molecules, we present predictions for an additional 27 molecules, including alkanes, alkenes, molecules with ring structures, and DNA nucleotide bases.

Learning thermodynamically constrained equations of state with uncertainty

Numerical simulations of high energy-density experiments require equation of state (EOS) models that relate a material’s thermodynamic state variables—specifically pressure, volume/density, energy, and temperature. EOS models are typically constructed using a semi-empirical parametric methodology, which assumes a physics-informed functional form with many tunable parameters calibrated using experimental/simulation data. Since there are inherent uncertainties in the calibration data (parametric uncertainty) and the assumed functional EOS form (model uncertainty), it is essential to perform uncertainty quantification (UQ) to improve confidence in EOS predictions. Model uncertainty is challenging for UQ studies since it requires exploring the space of all possible physically consistent functional forms. Thus, it is often neglected in favor of parametric uncertainty, which is easier to quantify without violating thermodynamic laws. This work presents a data-driven machine learning approach to constructing EOS models that naturally captures model uncertainty while satisfying the necessary thermodynamic consistency and stability constraints. We propose a novel framework based on physics-informed Gaussian process regression (GPR) that automatically captures total uncertainty in the EOS and can be jointly trained on both simulation and experimental data sources. A GPR model for the shock Hugoniot is derived, and its uncertainties are quantified using the proposed framework. We apply the proposed model to learn the EOS for the diamond solid state of carbon using both density functional theory data and experimental shock Hugoniot data to train the model and show that the prediction uncertainty is reduced by considering thermodynamic constraints.

Deep Learning-Based Prediction of Stress and Strain Maps in Arterial Walls for Improved Cardiovascular Risk Assessment

Conducting computational stress-strain analysis using finite element methods (FEM) is a common approach when dealing with the complex geometries of atherosclerosis, which is a leading cause of global mortality and complex cardiovascular disease. The considerable expense linked to FEM analysis encourages the substitution of FEM with a considerably faster data-driven machine learning (ML) approach. This study investigated the potential of end-to-end deep learning tools as a more effective substitute for FEM in predicting stress-strain fields within 2D cross sections of arterial walls. We first proposed a U-Net-based fully convolutional neural network (CNN) to predict the von Mises stress and strain distribution based on the spatial arrangement of calcification within arterial wall cross-sections. Further, we developed a conditional generative adversarial network (cGAN) to enhance, particularly from the perceptual perspective, the prediction accuracy of stress and strain field maps for arterial walls with various calcification quantities and spatial configurations. On top of U-Net and cGAN, we also proposed their ensemble approaches to improve the prediction accuracy of field maps further. Our dataset, consisting of input and output images, was generated by implementing boundary conditions and extracting stress-strain field maps. The trained U-Net models can accurately predict von Mises stress and strain fields, with structural similarity index scores (SSIM) of 0.854 and 0.830 and mean squared errors of 0.017 and 0.018 for stress and strain, respectively, on a reserved test set. Meanwhile, the cGAN models in a combination of ensemble and transfer learning techniques demonstrate high accuracy in predicting von Mises stress and strain fields, as evidenced by SSIM scores of 0.890 for stress and 0.803 for strain. Additionally, mean squared errors of 0.008 for stress and 0.017 for strain further support the model’s performance on a designated test set. Overall, this study developed a surrogate model for finite element analysis, which can accurately and efficiently predict stress-strain fields of arterial walls regardless of complex geometries and boundary conditions.

Battery Modeling: Fusing Equivalent Circuit Models with Data-Driven Surrogate Modeling

Electrochemical impedance spectroscopy (EIS) is widely used to characterize electrochemical energy conversion systems. The traditional analysis with equivalent circuit models (ECM) has recently been augmented by a transform based distribution of relaxation times (DRT) analysis which allows one to reduce the ambiguity in the construction of ECMs and thus overfitting. Experimentally determined ECM parameters vary depending on operating conditions and the lifetime history of battery usage. Here we focus on State of Health (SOH) and State of Charge (SOC) as a basis for operando diagnosis and functionality optimization in the setting of fast-charging. Within pure ECM approaches, aging effects can only be represented to a limited extent, as aging is related to a variety of different factors whose impact on cell impedance are not sufficiently understood, yet. The highly complex interplay of factors motivates the development of data-driven Machine Learning (ML) models as a basis for future battery management systems. We present ML enabled ECMs based on experimental impedance analyses and a data-driven ML approach that computationally samples an abstract target space for classification and recognition of cells at vastly different SOC/SOH combinations.

Random forest classification as a tool in epidemiological modelling: Identification of farm-specific characteristics relevant for the occurrence of Fasciola hepatica on German dairy farms.

Fasciola hepatica is an internal parasite of both human and veterinary relevance. In order to control fasciolosis, a multitude of attempts to predict the risk of infection such as risk maps or forecasting models have been developed. These attempts mainly focused on the influence of geo-climatic and meteorological features. Predicting bovine fasciolosis on farm level taking into account farm-specific settings yet remains challenging. In the present study, a new methodology for this purpose, a data-driven machine learning approach using a random forest classification algorithm was applied to a cross-sectional data set of farm characteristics, management regimes, and farmer aspects within two structurally different dairying regions in Germany in order to identify factors relevant for the occurrence of F. hepatica that could predict farm-level bulk tank milk positivity. The resulting models identified farm-specific key aspects in regard to the presence of F. hepatica. In study region North, farm-level production parameters (farm-level milk yield, farm-level milk fat, farm-level milk protein), leg hygiene, body condition (prevalence of overconditioned and underconditioned cows, respectively) and pasture access were identified as features relevant in regard to farm-level F. hepatica positivity. In study region South, pasture access together with farm-level lameness prevalence, farm-level prevalence of hock lesions, herd size, parity, and farm-level milk fat appeared to be important covariates. The stratification of the analysis by study region allows for the extrapolation of the results to similar settings of dairy husbandry. The local, region-specific modelling of F. hepatica presence in this work contributes to the understanding of on-farm aspects of F. hepatica appearance. The applied technique represents a novel approach in this context to model epidemiological data on fasciolosis which allows for the identification of farms at risk and together with additional findings in regard to the epidemiology of fasciolosis, can facilitate risk assessment and deepen our understanding of on-farm drivers of the occurrence of F. hepatica.

Unlocking Insights: Analysing COVID-19 Lockdown Policies and Mobility Data in Victoria, Australia, through a Data-Driven Machine Learning Approach

The state of Victoria, Australia, implemented one of the world’s most prolonged cumulative lockdowns in 2020 and 2021. Although lockdowns have proven effective in managing COVID-19 worldwide, this approach faced challenges in containing the rising infection in Victoria. This study evaluates the effects of short-term (less than 60 days) and long-term (more than 60 days) lockdowns on public mobility and the effectiveness of various social restriction measures within these periods. The aim is to understand the complexities of pandemic management by examining various measures over different lockdown durations, thereby contributing to more effective COVID-19 containment methods. Using restriction policy, community mobility, and COVID-19 data, a machine-learning-based simulation model was proposed, incorporating analysis of correlation, infection doubling time, and effective lockdown date. The model result highlights the significant impact of public event cancellations in preventing COVID-19 infection during short- and long-term lockdowns and the importance of international travel controls in long-term lockdowns. The effectiveness of social restriction was found to decrease significantly with the transition from short to long lockdowns, characterised by increased visits to public places and increased use of public transport, which may be associated with an increase in the effective reproduction number (Rt) and infected cases.

Integrated machine learning methods with oversampling technique for regional suitability prediction of waste-to-energy incineration projects

China's tiered strategy to enhance county-level waste incineration for energy aligns with the sustainable development goals (SDGs), emphasizing the need for comprehensive assessments of waste-to-energy (WtE) plant suitability. Traditional assessment methodologies face challenges, particularly in suggesting innovative site alternatives, adapting to new data sets, and their dependence on strict assumptions. This study introduced enhancements in three pivotal dimensions. Methodologically, it leverages data-driven machine learning (ML) approaches to capture the complex relationships essential for site selection, reducing dependency on strict assumptions. In terms of predictive performance, the integration of oversampling with stacked ensemble models enhances the diversity and generalizability of ML models. The area under curve (AUC) scores from four ML models, enhanced by the oversampled dataset, demonstrated significant improvements compared to the original dataset. The stacking model excelled, achieving a score of 92%. It also led in overall Precision and Recall, reaching 85.2% and 85.08% respectively. Nevertheless, a noticeable discrepancy existed in Precision and Recall for positive classes. The stacking model topped Precision scores at 83.1%, followed by eXtreme Gradient Boosting (XGBoost) (82.61%). In terms of Recall, XGBoost recorded the lowest at 85.07%, while the other three classifiers all marked 88.06%. From an industry applicability standpoint, the stacking model provides innovative location alternatives and demonstrates adaptability in Hunan province, offering a reusable tool for WtE location. In conclusion, this study not only enhances the methodological aspects of WtE site selection but also provides practical and adaptable solutions, contributing positively to sustainable waste management practices.

Early Prediction of Impending Exertional Heat Stroke With Wearable Multimodal Sensing and Anomaly Detection.

We employed wearable multimodal sensing (heart rate and triaxial accelerometry) with machine learning to enable early prediction of impending exertional heat stroke (EHS). US Army Rangers and Combat Engineers (N = 2,102) were instrumented while participating in rigorous 7-mile and 12-mile loaded rucksack timed marches. There were three EHS cases, and data from 478 Rangers were analyzed for model building and controls. The data-driven machine learning approach incorporated estimates of physiological strain (heart rate) and physical stress (estimated metabolic rate) trajectories, followed by reconstruction to obtain compressed representations which then fed into anomaly detection for EHS prediction. Impending EHS was predicted from 33 to 69 min before collapse. These findings demonstrate that low dimensional physiological stress to strain patterns with machine learning anomaly detection enables early prediction of impending EHS which will allow interventions that minimize or avoid pathophysiological sequelae. We describe how our approach can be expanded to other physical activities and enhanced with novel sensors.

A data-driven ensemble machine learning approach for predicting the mechanical strength of 3D printed orthopaedic bone screws

A data-driven ensemble machine learning approach for predicting the mechanical strength of 3D printed orthopaedic bone screws