Physics-based Methods Research Articles

From data to insight, enhancing structural health monitoring using physics-informed machine learning and advanced data collection methods

Abstract Owing to recent advancements in sensor technology, data mining, Machine Learning (ML) and cloud computation, Structural Health Monitoring (SHM) based on a data-driven approach has gained more popularity and interest. The data-driven methodology has proved to be more efficient and robust compared with traditional physics-based methods. The past decade has witnessed remarkable progress in ML, especially in the field of Deep Learning (DL) which are effective in many tasks and has achieved state-of-the-art results in various engineering domains. In the same manner, DL has also revolutionized SHM technology by improving the effectiveness and efficiency of models, as well as enhancing safety and reliability. To some extent, it has also paved the way for implementing SHM in real-world complex civil and mechanical infrastructures. However, despite all the success, DL has intrinsic limitations such as its massive-labelled data Requirement, inability to generate consistent results and lack of generalizability to out-of-sample scenarios. Conversely, in SHM, the lack of data corresponding to a different state of the structure is still a challenging task. Recent development in physics-informed ML methods has provided an opportunity to resolve these challenges in which limited-noisy data and mathematical models are integrated through ML algorithms. This method automatically satisfies physical invariants providing better accuracy and improved generalization. This manuscript presents the sate-of-the-art review of prevailing ML methods for efficient damage inspection, discuss their limitations, and explains the diverse applications and benefits of physics-informed ML in the SHM setting. Moreover, the latest data extraction strategy and the internet of things (IoT) that support the present data-driven methods and SHM are also briefly discussed in the last section.

Predicting gas diffusion layer flow information in proton exchange membrane fuel cells from cross-sectional data using deep learning methods

Obtaining transient flow field information of gas diffusion layers (GDLs) is a crucial issue for improving the performance of proton exchange membrane fuel cells (PEMFCs). While physically-based methods, such as effective medium theory and solving partial differential equations, can be utilized to calculate fluid flow in GDL, the computational cost associated with such methods remains substantial. In this study, a 3D multiphase flow dataset of GDLs is obtained using fluid of volume (VOF) calculation, and the 3D data is sliced. The resulting cross-sectional data are used to establish a convolutional neural network (CNN) model based on two-dimensional (2D) data for predicting fluid flow in GDLs. The reliability of predicting the saturation information of 3D GDLs using 2D cross-sectional data is verified and discussed. Subsequently, a comparison is made between the accuracy and computational cost of predicting the saturation of 3D GDLs using 2D and 3D CNN models. The results indicate that using 25 cross-sectional images provide accurate predictions of GDL saturation in 3D. When using cross-sectional images as inputs, 100 × 100 images are found to be more representative than 50 × 50 images, resulting in an average increase of 1.86% and 13.36% in R2 and RMSE, respectively. While the 3D CNN model outperforms the 2D CNN model in predicting GDL saturation in 3D by only 0.62% in terms of R2 score, its computational cost is two orders of magnitude higher. These findings suggest that 2D GDL cross-sectional images can also be used for predicting 3D GDL flow information and effectively reducing computational costs. The findings of this study provide a profound insight into the GDL flow phenomena and contribute to the development of more efficient and accurate fuel cell models.

Estimation of elastic bandgaps in metastructures: A comparison of physics-based and data-driven approaches

Metastructures are an emerging solution for applications in aerospace, industry, and robotics, as they inhibit the propagation of elastic waves within a specific frequency range called the “bandgap”. Accurate estimation of bandgaps is crucial for optimizing metastructures for specific purposes. Two approaches have been traditionally used: physics-based modeling, which requires precise characterization of the unit cell’s physical properties, and data-driven methods based on steady-state dynamic response. This study compares the effectiveness of data-driven methods (Component Mode Synthesis and FRF-Based Substructuring) with traditional physics-based methods for identifying bandgaps in multi-unit cell metastructures. We also validate the identified bandgaps using experimental reading-based methods. Our goal is to determine a more efficient and accurate approach for identifying bandgaps in metastructures, with potential implications across various fields.

A comparison of ultrasonic temperature monitoring using machine learning and physics-based methods for high-cycle thermal fatigue monitoring

Failure of pipe network components in so-called mixing zones due to high-cycle thermal fatigue (HCTF) can occur within nuclear power plants where fluids of different thermal and hydraulic properties interact. Given that the consequences of such failures are potentially deadly, a method to monitor HCTF non-invasively in real-time is expected to be of great use. This method may be realised by a technique to determine the inaccessible temperature distribution of a component since thermal gradients drive HCTF. Previous work showed that a physics-based method called the inverse thermal modelling (ITM) method can obtain the temperature distribution from external temperature and ultrasonic time of flight (TOF) measurements. This study investigated whether the long-short-term memory (LSTM) machine learning architecture could be a faster alternative to the ITM method for data inversion. On experimental data, a 25-member ensemble of LSTM networks achieved an ensemble median root mean square error (RMSE) of 1.04°C and an ensemble median mean error of 0.194°C (both relative to a resistance temperature device measurement). These values are similar to the ITM method which achieved a RMSE of 1.04°C and a mean error of 0.196°C. The single LSTM network and the ITM method achieved a computation-to-real-world time ratio of 0.008% and 14%, respectively demonstrating that both methods can invert data in real-time. Simulation studies revealed that LSTM performance is sensitive to small differences between the training and real-world parameters leading to unacceptable errors. However, these errors can be detected via an ensemble of independent networks and, corrected by simply adding a correction factor to the TOF prior to being input into the networks. The results show that LSTM has the potential to be an alternative to the ITM method; however, the authors favour ITM for temperature distribution monitoring given its interpretability.

Computational Methods in Immunology and Vaccinology: Design and Development of Antibodies and Immunogens.

The design of new biomolecules able to harness immune mechanisms for the treatment of diseases is a prime challenge for computational and simulative approaches. For instance, in recent years, antibodies have emerged as an important class of therapeutics against a spectrum of pathologies. In cancer, immune-inspired approaches are witnessing a surge thanks to a better understanding of tumor-associated antigens and the mechanisms of their engagement or evasion from the human immune system. Here, we provide a summary of the main state-of-the-art computational approaches that are used to design antibodies and antigens, and in parallel, we review key methodologies for epitope identification for both B- and T-cell mediated responses. A special focus is devoted to the description of structure- and physics-based models, privileged over purely sequence-based approaches. We discuss the implications of novel methods in engineering biomolecules with tailored immunological properties for possible therapeutic uses. Finally, we highlight the extraordinary challenges and opportunities presented by the possible integration of structure- and physics-based methods with emerging Artificial Intelligence technologies for the prediction and design of novel antigens, epitopes, and antibodies.

Δ-Learning applied to coarse-grained homogeneous liquids.

Coarse-grained molecular dynamics (CGMD) simulations address lengthscales and timescales that are critical to many chemical and material applications. Nevertheless, contemporary CGMD modeling is relatively bespoke and there are no black-box CGMD methodologies available that could play a comparable role in discovery applications that density functional theory plays for electronic structure. This gap might be filled by machine learning (ML)-based CGMD potentials that simplify model development, but these methods are still in their early stages and have yet to demonstrate a significant advantage over existing physics-based CGMD methods. Here, we explore the potential of Δ-learning models to leverage the advantages of these two approaches. This is implemented by using ML-based potentials to learn the difference between the target CGMD variable and the predictions of physics-based potentials. The Δ-models are benchmarked against the baseline models in reproducing on-target and off-target atomistic properties as a function of CG resolution, mapping operator, and system topology. The Δ-models outperform the reference ML-only CGMD models in nearly all scenarios. In several cases, the ML-only models manage to minimize training errors while still producing qualitatively incorrect dynamics, which is corrected by the Δ-models. Given their negligible added cost, Δ-models provide essentially free gains over their ML-only counterparts. Nevertheless, an unexpected finding is that neither the Δ-learning models nor the ML-only models significantly outperform the elementary pairwise models in reproducing atomistic properties. This fundamental failure is attributed to the relatively large irreducible force errors associated with coarse-graining that produces little benefit from using more complex potentials.

Recent Advances and Applications of Machine Learning in Experimental Solid Mechanics: A Review

Abstract For many decades, experimental solid mechanics has played a crucial role in characterizing and understanding the mechanical properties of natural and novel artificial materials. Recent advances in machine learning (ML) provide new opportunities for the field, including experimental design, data analysis, uncertainty quantification, and inverse problems. As the number of papers published in recent years in this emerging field is growing exponentially, it is timely to conduct a comprehensive and up-to-date review of recent ML applications in experimental solid mechanics. Here, we first provide an overview of common ML algorithms and terminologies that are pertinent to this review, with emphasis placed on physics-informed and physics-based ML methods. Then, we provide thorough coverage of recent ML applications in traditional and emerging areas of experimental mechanics, including fracture mechanics, biomechanics, nano- and micromechanics, architected materials, and two-dimensional materials. Finally, we highlight some current challenges of applying ML to multimodality and multifidelity experimental datasets, quantifying the uncertainty of ML predictions, and proposing several future research directions. This review aims to provide valuable insights into the use of ML methods and a variety of examples for researchers in solid mechanics to integrate into their experiments.

A Deep Transfer Learning Network for Structural Condition Identification with Limited Real-World Training Data

Structural condition identification based on monitoring data is important for automatic civil infrastructure asset management. Nevertheless, the monitoring data are almost always insufficient because the real-time monitoring data of a structure only reflect a limited number of structural conditions, while the number of possible structural conditions is infinite. With insufficient monitoring data, the identification performance may significantly degrade. This study aims to tackle this challenge by proposing a deep transfer learning (TL) approach for structural condition identification. It effectively integrates physics-based and data-driven methods by generating various training data based on the calibrated finite element (FE) model, pretraining a deep learning (DL) network, and transferring its embedded knowledge to the real monitoring/testing domain. Its performance is demonstrated in a challenging case, vibration-based condition identification of steel frame structures with bolted connection damage. First, disparate subsets of test data are used as training data, and the identification accuracy of the whole dataset is evaluated. The results demonstrate that the proposed approach can achieve high identification accuracy with limited types of training data, with the identification accuracy increasing up to 8.57%. Second, numerical simulation data are used as training data, and then different TL strategies and different DL architectures are compared on the performance of structural condition identification. The results show that even though the training data are from a different domain and with different types of labels, intrinsic physics can be learned through the pretraining process, and the TL results can be clearly improved, with the identification accuracy increasing from 81.8% to 89.1%. The comparative studies show that SHMnet with three convolutional layers stands out as the pretraining DL architecture, with 21.8% and 25.5% higher identification accuracy values over the other two networks, VGGNet-16 and ResNet-18. The findings of this study advance the potential application of the proposed approach towards expert-level condition identification based on limited real-world training data.

Comparison between Machine Learning and Physical Models Applied to the Evaluation of Co-Seismic Landslide Hazard

A comparative methodology between advanced statistical tools and physical-based methods is carried out to ensure their reliability and objectivity for the evaluation of co-seismic landslide hazard maps. To do this, an inventory of landslides induced by the 2011 Lorca earthquake is used to highlight the usefulness of these methods to improve earthquake-induced landslide hazard analyses. Various statistical models, such as logistic regression, random forest, artificial neural network, and support vector machine, have been employed for co-seismic landslide susceptibility mapping. The results demonstrate that machine learning techniques using principal components (especially, artificial neural network and support vector machine) yield better results compared to other models. In particular, random forest shows poor results. Artificial neural network and support vector machine approaches are compared to the results of physical-based methods in the same area, suggesting that machine learning methods can provide better results for developing co-seismic landslide susceptibility maps. The application of different advanced statistical models shows the need for validation with an actual inventory of co-seismic landslides to ensure reliability and objectivity. In addition, statistical methods require a great amount of data. The results establish effective land planning and hazard management strategies in seismic areas to minimize the damage of future co-seismic landslides.

Physics-informed machine learning in prognostics and health management: State of the art and challenges

Prognostics and health management (PHM) plays a constructive role in the equipment’s entire life health service. It has long benefited from intensive research into physics modeling and machine learning methods. However, in practice, the existing solutions often encounter difficulties caused by sparse data & incomplete system failure knowledge. Pure machine learning or physics-based methods can sometimes be infeasible in such situations. As a result, there has been a growing interest in developing physics-informed machine learning (PIML) models which allow incorporating different forms of physics knowledge at different positions of the machine learning pipeline. This combination provides significant assistance for detection, diagnostic, and prognostics. However, to the best of our knowledge, the bibliometrics analyses and the comprehensive review of the existing research concerning PIML in PHM remain vacant. Our review is therefore dedicated to filling these gaps. We synthesize the concept of PIML in PHM, and propose a taxonomy of PIML approaches from the perspective of “Expression forms of informed knowledge” and “Knowledge informed methods”. The findings and discussions presented in this paper enable us to clarify the current state of the art and the emerging opportunities of PIML approaches, especially for building PHM systems that can work under the “small data and scarce physics knowledge” paradigm.

A Space-Time Partial Differential Equation Based Physics-Guided Neural Network for Sea Surface Temperature Prediction

Sea surface temperature (SST) prediction has attracted increasing attention, due to its crucial role in understanding the Earth’s climate and ocean system. Existing SST prediction methods are typically based on either physics-based numerical methods or data-driven methods. Physics-based numerical methods rely on marine physics equations and have stable and explicable outputs, while data-driven methods are flexible in adapting to data and are capable of detecting unexpected patterns. We believe that these two types of method are complementary to each other, and their combination can potentially achieve better performances. In this paper, a space-time partial differential equation (PDE) is employed to form a novel physics-based deep learning framework, named the space-time PDE-guided neural network (STPDE-Net), to predict daily SST. Comprehensive experiments for SST prediction were conducted, and the results proved that our method could outperform the traditional finite-difference forecast method and several state-of-the-art deep learning and physics-guided deep learning methods.

Machine learning simulation of one-dimensional deterministic water wave propagation

Deterministic phase-resolved prediction of the evolution of surface gravity waves in water is challenging due to their complex spatio-temporal dynamics. Physics-based methods of varying complexity are available, but the conflicting objectives of numerical efficiency and accuracy impede real-time wave prediction. Data-driven methods may be able to overcome this challenge by using training data generated by complex numerical methods. This work explores the potential of a machine learning (ML) approach based on a fully convolutional encoder–decoder architecture for the efficient and accurate prediction of water waves. The high-order spectral (HOS) method forms the foundation for the generation of the training data. The HOS method is applied for different, consecutive orders of nonlinearity starting from first order up to fourth order. The JONSWAP wave energy spectrum serves as the basis for modeling the one-dimensional irregular sea states. The overall objective of this work is to evaluate whether the complex non-linear physical processes can be identified and learned by the ML approach. The trained ML flow mapper is used to perform time integration of an initial sea state. The results indicate that the proposed ML approach is able to reproduce the distinctive physical processes of the different orders of nonlinearities. It is shown that the ML approach enables fast and accurate predictions of one-dimensional waves over a time horizon that spans multiple peak periods.

Skilful nowcasting of extreme precipitation with NowcastNet

Extreme precipitation is a considerable contributor to meteorological disasters and there is a great need to mitigate its socioeconomic effects through skilful nowcasting that has high resolution, long lead times and local details1–3. Current methods are subject to blur, dissipation, intensity or location errors, with physics-based numerical methods struggling to capture pivotal chaotic dynamics such as convective initiation4 and data-driven learning methods failing to obey intrinsic physical laws such as advective conservation5. We present NowcastNet, a nonlinear nowcasting model for extreme precipitation that unifies physical-evolution schemes and conditional-learning methods into a neural-network framework with end-to-end forecast error optimization. On the basis of radar observations from the USA and China, our model produces physically plausible precipitation nowcasts with sharp multiscale patterns over regions of 2,048 km × 2,048 km and with lead times of up to 3 h. In a systematic evaluation by 62 professional meteorologists from across China, our model ranks first in 71% of cases against the leading methods. NowcastNet provides skilful forecasts at light-to-heavy rain rates, particularly for extreme-precipitation events accompanied by advective or convective processes that were previously considered intractable.

Electron Beam Irradiation to Control Rhizoctonia solani in Potato

This study focuses on the influence of pre-planting irradiation on the development, health, and yield of seed potatoes infected with Rhizoctonia solani. The research was prompted by the need to ensure crop security and sustainability in the modern-day environment, which calls into question the future sufficiency of crop yields. Considering that the focus has shifted to non-chemical methods of crop treatment at all plant development stages in response to more stringent regulations governing potato production, it is particularly important to refine physics-based methods to suppress fungal diseases caused by Rhizoctonia solani. Irradiation of tubers with 20–150 Gy inhibited the potato development phases and the doses exceeding 150 Gy completely suppressed the potato sprouting. Doses ranging from 20 Gy to 100 Gy decreased the quantity of large tubers by 10–20% on average while the number of medium and small tubers increased by 5–15% and 3–10%, respectively. Irradiation of seed potatoes also decreased the sclerotia and non-sclerotia forms of diseases caused by Rhizoctonia solani in the harvested tubers. It was found that 1 MeV electron irradiation with doses ranging from 20 Gy to 30 Gy is the most efficient for the pre-planting treatment of seed potatoes since the penetration of low-energy accelerated electrons into the upper layers of potato tubers ensures the suppression of diseases caused by Rhizoctonia solani by at least 10% from the value of non-irradiated samples and prevents the reduction of total yield allowing for a maximum of 25% loss.

Real-Time Forecasting of Subsurface Inclusion Defects for Continuous Casting Slabs: A Data-Driven Comparative Study

Subsurface inclusions are one of the most common defects that affect the inner quality of continuous casting slabs. This increases the defects in the final products and increases the complexity of the hot charge rolling process and may even cause breakout accidents. The defects are, however, hard to detect online by traditional mechanism-model-based and physics-based methods. In the present paper, a comparative study is carried out based on data-driven methods, which are only sporadically discussed in the literature. As a further contribution, a scatter-regularized kernel discriminative least squares (SR-KDLS) model and a stacked defect-related autoencoder back propagation neural network (SDAE-BPNN) model are developed to improve the forecasting performance. The scatter-regularized kernel discriminative least squares is designed as a coherent framework to directly provide forecasting information instead of low-dimensional embeddings. The stacked defect-related autoencoder back propagation neural network extracts deep defect-related features layer by layer for a higher feasibility and accuracy. The feasibility and efficiency of the data-driven methods are demonstrated through case studies based on a real-life continuous casting process, where the imbalance degree drastically vary in different categories, showing that the defects are timely (within 0.01 ms) and accurately forecasted. Moreover, experiments illustrate the merits of the developed scatter-regularized kernel discriminative least squares and stacked defect-related autoencoder back propagation neural network methods regarding the computational burden; the F1 scores of the developed methods are clearly higher than common methods.

A Hybrid Approach of Data-driven and Physics-based Methods for Estimation and Prediction of Fatigue Crack Growth

Lamb-wave-based nondestructive testing and evaluation (NDT/E) methods have drawn much attention due to their potential to inspect plate-like structures in a variety of industrial applications. To estimate and/or predict fatigue crack growth, many research efforts have been made to develop data-driven or physics-based methods. Data-driven methods show high predictive capability without the need for physical domain knowledge; however, fewer data can lead to overfitting in the results. On the other hand, physics-based methods can provide reliable results without the need for measured data; however, small amounts of physical information can worsen their predictive capability. In real applications, both the measurable data and the physical information of systems may be considerably limited; it is thus challenging to estimate and/or predict the crack length using either the data-driven or physics-based method alone. To make use of the advantages and minimize the disadvantages of each method, the work outlined in this paper aims to develop a hybrid approach that combines the data-driven and the physics-based methods for estimation and prediction of fatigue crack growth with and without Lamb wave signals. First, with Lamb wave signals, a data-driven method based on signal processing and the random forest model can be used estimate crack lengths. Second, in the absence of Lamb wave signals, a physics-based method based on an ensemble prognostics approach and Walker’s equation can be used to predict crack lengths with the help of the previously estimated crack lengths. To demonstrate the validity of the proposed approach, a case study is presented using datasets provided in the 2019 PHM Conference Data Challenge by the PHM Society. The case study confirms that the proposed method shows high accuracy; the RMSEs for specimens T7 and T8 are calculated as 0.2021 and 0.551, respectively. A penalty score is calculated as 7.63, this result led to a 2nd place finish in the Data Challenge. To the best of the authors’ knowledge, this is the first attempt to propose a hybrid approach for estimation and prediction of fatigue crack growth.

Systems Integration Framework for Hybrid-Electric Commuter and Regional Aircraft

System integration is one of the key challenges to bringing future hybrid-electric and all-electric aircraft into the market. In addition, retrofitting and redesigning existing aircraft are potential paths toward achieving hybrid and all-electric flight, which are even more challenging goals from a system integration perspective. Therefore, integration tools that bridge the gap between the aircraft and the subsystem level need to be developed for use in the conceptual design stage to address current system integration challenges, such as the use of space, the share between propulsive and secondary power, required level of electrification, safety, and thermal management. This paper presents a multidisciplinary design analysis (MDA) framework that integrates aircraft and subsystem sizing tools. In addition, this paper includes improved physics-based subsystem sizing methods that are also applicable to smaller, commuter, or regional aircraft. The capabilities of the developed framework and tools are presented for a case study covering the redesign of the DO-228 with a hybrid-electric propulsion system in combination with the electrification of its systems architecture and different subsystem technologies.

A Novel Hybrid Physics-Based and Data-Driven Approach for Degradation Trajectory Prediction in Li-Ion Batteries

Lithium-ion batteries have been widely used in electric vehicles. To ensure safety and reliability, accurate prediction of the battery’s future degradation trajectory is critical. However, early prediction capability and adaptive prediction capability under various battery aging conditions remain two main challenges. Either physics-based or data-driven methods have their advantages and limitations. In this study, a novel hybrid method that combines the physics-based and data-driven approaches is proposed to achieve early prediction of the battery capacity degradation trajectory. This framework consists of three steps. First, to improve the generality of the method, a hybrid feature is extracted using an electrochemical model and measured voltage data. Second, the clustering algorithm is adopted to divide battery degradation data into different clusters, and the data augmentation technique is used to enrich the training dataset. Finally, the training dataset in each cluster is used to train the sequence-to-sequence deep neural network, and the future degradation trajectory can be predicted. The proposed method provides accurate predictions using only 20% training data, and it has strong robustness under noisy input. Validation results under different aging conditions show that the mean absolute percentage errors of capacity degradation trajectory and remaining useable cycle life are below 2.5% and 6.5%, respectively.

A review of machine learning models and influential factors for estimating evapotranspiration using remote sensing and ground-based data

In the era of water scarcity and severe droughts, the accurate estimation of evapotranspiration (ET) is crucial for the efficient management of water resources, understanding hydrological and ecological processes, and comprehending the relationships between the atmosphere, hydrosphere, and biosphere. ET is a complex phenomenon influenced by a set of biophysical and environmental factors. Its estimation becomes more complicated in heterogeneous environments, demanding detailed data and accurate model calibration. Combining remote sensing imagery and machine learning (ML) models has provided a considerable capacity for estimating ET, which relaxes a number of assumptions and requires less data than traditional approaches. Satellite imagery provides influential variables for ET estimation using ML models. Nevertheless, a growing number of ML models and emerging satellite imagery has opened up a wide and complex potential before researchers. While previous studies have reviewed physical-based methods for ET estimation, this paper offers a recent decade review of the progress, challenges, and opportunities provided by the RS and ML models for the ET estimation and future outlook.

Modeling Methodology Based on Fast and Refined Neural Networks for Non-Isolated DC–DC Converters With Configurable Parameter Settings

Compared with conventional physics-based methods, e.g., analytical modeling and numerical modeling, data-driven methods can extract input-to-output relationships from the data without much prior knowledge of the physical system, thus showing great potential in modeling power electronics (PE) converters with complex switching behaviors and configurable parameter settings. Previous data-driven PE circuit modeling approaches are mostly based on sequential neural networks, and their execution speed suffers from large sequential lengths due to a high sampling rate for high modeling accuracy. Moreover, modeling of refined singular ripples is missing and configurable parameter settings are not available in these data-driven modeling approaches. To address the above-mentioned issues, this paper proposes a hybrid physics-informed machine learning (ML) method to model the non-isolated DC-DC converters. The approach empirically decomposes the output signals into transient large signals and periodic small signals. For transient large signals, a fully-connected neural network (NN) is used to map circuit parameters with system characteristics, such that configurable circuit parameter settings are allowed. For periodic signals, a long short-time memory (LSTM) network together with convolutional neural network (CNN) is used to accelerate the simulation by predicting signal features in the compressed latent space. A buck converter with configurable parameter settings is modeled by the proposed hybrid physics-informed ML method. Periodic ripples are successfully generated, while execution speed is about 10 times faster than that of conventional numerical methods.