Physics-based Approaches Research Articles

Degradation Prediction of Semiconductor Lasers Using Conditional Variational Autoencoder

Semiconductor lasers have been rapidly evolving to meet the demands of next-generation optical networks. This imposes much more stringent requirements on the laser reliability, which are dominated by degradation mechanisms (e.g., sudden degradation) limiting the semiconductor laser lifetime. Physics-based approaches are often used to characterize the degradation behavior analytically, yet explicit domain knowledge and accurate mathematical models are required. Building such models can be very challenging due to a lack of a full understanding of the complex physical processes inducing the degradation under various operating conditions. To overcome the aforementioned limitations, we propose a new data-driven approach, extracting useful insights from the operational monitored data to predict the degradation trend without requiring any specific knowledge or using any physical model. The proposed approach is based on an unsupervised technique, a conditional variational autoencoder, and validated using vertical-cavity surface-emitting laser (VCSEL) and tunable edge emitting laser reliability data. The experimental results confirm that our model (i) achieves a good degradation prediction and generalization performance by yielding an F1 score of 95.3%, (ii) outperforms several baseline ML based anomaly detection techniques, and (iii) helps to shorten the aging tests by early predicting the failed devices before the end of the test and thereby saving costs

Micromechanical damping analysis of composite materials using linear viscoelasticity vs the strain energy approach

The phenomenological strain energy approach for predicting composite damping properties is widely used because of its simplicity. The composite dissipated energy is approximated as the volume average of the local (constituent) loss factor times the local strain energy. Thus, any micromechanics method (e.g., finite element or the high-fidelity generalized method of cells used herein) that can predict the local fields (and thus the strain energy) within a composite can be used to approximate the effective damping properties. In this paper, two more physics-based approaches, the correspondence principle and explicit viscoelastic modeling, are implemented within the two-scale high-fidelity generalized method of cells micromechanics theory and compared with the widely-used strain energy approach. It is shown that all three approaches provide quite similar predictions for the effective damping properties of unidirectional composites. Finally, the presented two-scale micromechanics theory is applied to predict the effective anisotropic damping properties of 2D and 3D woven composites.

AI-based approach to predict nozzle instability in Pelton turbine model test.

Over past experience on reduced scale physical models of Pelton turbines, it has been noticed that a few models exhibited disturbed performance curves over a given domain of speed factor and discharge factor under specific test head values, while the same performance curves were smooth at other test heads. Cross-checking of different model component combinations helped identify that those disturbances occur with specific nozzles independently from other model components. The experienced disturbances have been understood to be consequences of instabilities attributed to the nozzle.Attempts to understand root causes and predict such instabilities with physics-based approaches first proved unsuccessful. The proposed research presents the analysis performed on an available model test results database to predict the occurrence of instability thanks to neural network classifiers, the so-called AI-based approach.A model test campaign including a nozzle known to be subject to instabilities but not part of the database used to train the classifier has been designed and conducted. The agreement between the operating domain with predicted disturbance and the operating domain with measured disturbance is finally discussed.

Real-Time Short-Term Pedestrian Trajectory Prediction Based on Gait Biomechanics.

The short-term prediction of a person’s trajectory during normal walking becomes necessary in many environments shared by humans and robots. Physics-based approaches based on Newton’s laws of motion seem best suited for short-term predictions, but the intrinsic properties of human walking conflict with the foundations of the basic kinematical models compromising their performance. In this paper, we propose a short-time prediction method based on gait biomechanics for real-time applications. This method relays on a single biomechanical variable, and it has a low computational burden, turning it into a feasible solution to implement in low-cost portable devices. We evaluate its performance from an experimental benchmark where several subjects walked steadily over straight and curved paths. With this approach, the results indicate a performance good enough to be applicable to a wide range of human–robot interaction applications.

Review of predictive maintenance algorithms applied to HVAC systems

Predictive maintenance is a preventive maintenance approach that is performed based on an online health assessment and allows for timely pre-failure interventions. It can diminish the cost of maintenance by reducing the frequency of maintenance as much as possible to avoid unplanned reactive maintenance, without incurring the costs associated with too frequent preventive maintenance. The main objective of predictive maintenance of heating, ventilation, and air conditioning (HVAC) systems is to predict when the HVAC equipment failure may occur. The benefits are numerous: planning of maintenance before the failure occurs, reduction of maintenance costs, and increased reliability. For this, the predictive maintenance of the HVAC systems is based on the historical data of the system for predicting the state of health of the system. The process of predictive maintenance application is composed of the Internet of Things (IoT) sensors that are installed inside the HVAC system, then the IoT platforms that help in collecting the signals coming from the sensors and converting them to existing databases. Afterward, the algorithms of application of predictive maintenance could be either knowledge-based approaches, physics-based approaches, or even data-driven-based approaches. A systematic literature review on the existing algorithms of HVAC predictive maintenance application is conducted in this paper to summarize the most used approach for predicting future failures in HVAC systems and to explain the benefits and limits of these algorithms.

A new 3-D finite-fault model of the Southwest Iceland bookshelf transform zone

SUMMARYThe largest earthquakes in Iceland occur in the South Iceland seismic zone (SISZ) and the Tjörnes fracture zone in the northeast. With the latter being primarily offshore, the seismic risk in Iceland is highest in the relatively densely populated SISZ. Past probabilistic seismic hazard assessment (PSHA) efforts in Iceland have, however, been based on statistical analyses of various historical earthquake catalogues and limited ground motion models (GMMs), all subject to varying types and degrees of uncertainties. Moreover, they relied on simplistic source descriptions and largely ignored that the unique ‘bookshelf’ strike-slip fault system of the SISZ extends along the plate margins towards the west and over the entire Reykjanes Peninsula Oblique Rift (RPOR) zone. Namely, the bookshelf fault system in Southwest Iceland is twice as long as previously thought and it dominates the strain release of transcurrent plate motion in Southwest Iceland, having potentially important implications for PSHA. In this study, therefore, we propose a new 3-D finite-fault model of the Southwest Iceland bookshelf transform zone. The model has been calibrated on the basis of first principles to the rate of transcurrent plate motions across the transform zone and constrained by the salient features of the fault system geometry as reported in the literature. We model the systematic spatial variability of the seismogenic potential along the zone by its provisional subdivision into six distinct zones. The fault system model allows both deterministic and random fault locations, with each realization completely specified in terms of the maximum expected magnitude of each fault, its maximum dimensions and its long-term slip rate. The variability of the model has been estimated through sensitivity analyses of its key parameters. The total seismic moment rates produced by the fault system model are completely consistent with those reported in the literature. The new model allows the derivation of simple but self-consistent zone-specific Gutenberg–Richter (GR) relationships, and the total long-term seismic activity predicted by the new 3-D fault system model effectively explains the historical earthquake catalogue of the SISZ–RPOR transform zone in Southwest Iceland. We are therefore confident that the model can serve as the foundation for future time-independent physics-based PSHA for Southwest Iceland. Moreover, the consistency and versatility of the model allows its application in conventional approaches to PSHA, which has the potential of bridging the gap between physics-based and conventional approaches to PSHA in Southwest Iceland. Such efforts will improve our understanding of the key elements that affect the hazard, thus improving the reliability of hazard estimates, with important practical implications for the optimized assessment of seismic risk.

Revisiting Theoretical Tools and Approaches for the Valorization of Recalcitrant Lignocellulosic Biomass to Value-Added Chemicals

Biorefinery processes for converting lignocellulosic biomass to fuels and chemicals proceed via an integrated series of steps. Biomass is first pretreated and deconstructed using chemical catalysts and/or enzymes to liberate sugar monomers and lignin fragments. Deconstruction is followed by a conversion step in which engineered host organisms assimilate the released sugar monomers and lignin fragments, and produce value-added fuels and chemicals. Over the past couple of decades, a significant amount of work has been done to develop innovative biomass deconstruction and conversion processes that efficiently solubilize biomass, separate lignin from the biomass, maximize yields of bioavailable sugars and lignin fragments and convert the majority of these carbon sources into fuels, commodity chemicals, and materials. Herein, we advocate that advanced in silico approaches provide a theoretical framework for developing efficient processes for lignocellulosic biomass valorization and maximizing yields of sugars and lignin fragments during deconstruction and fuel and chemical titers during conversion. This manuscript surveys the latest developments in lignocellulosic biomass valorization with special attention given to highlighting computational approaches used in process optimization for lignocellulose pretreatment; enzyme engineering for enhanced saccharification and delignification; and prediction of the genome modification necessary for desired pathway fine-tuning to upgrade products from biomass deconstruction into value-added products. Physics-based modeling approaches such as density functional theory calculations and molecular dynamics simulations have been most impactful in studies aimed at exploring the molecular level details of solvent-biomass interactions, reaction mechanisms occurring in biomass-solvent systems, and the catalytic mechanisms and engineering of enzymes involved in biomass degradation. More recently, with ever increasing amounts of data from, for example, advanced mutli-omics experiments, machine learning approaches have begun to make important contributions in synthetic biology and optimization of metabolic pathways for production of biofuels and chemicals.

Estimation of fatigue crack initiation and growth in engineering product development using a digital twin approach

A digital twin is a computational model in cyberspace that is used to support engineering decisions. Maintaining high predictive capability of a digital twin model is of great concern to the engineers who make design decisions at the early stages of product development. In the work described in this paper, the predictive capability of the digital twin approach is improved by considering uncertainties in manufacturing and test conditions. The proposed digital twin approach can be used in a variety of product development settings. The proposed idea takes advantage of hybrid digital twin approaches, using both data-driven and physics-based approaches. The proposed approach is based on two techniques; (i) statistical model calibration and (ii) probabilistic element updating. In statistical model calibration, statistical parameters of input variables are estimated. Further, probabilistic analysis using estimated statistical parameters can predict possible critical elements. In probabilistic element updating procedures, the possible crack initiation and growth element is updated. The validity of the proposed method is demonstrated using a case study of an automotive sub-frame fatigue test. From the results, we conclude that the proposed digital twin approach can accurately estimate crack initiation and growth of an automotive structure under uncertain loading conditions and material properties.

Computational Assessment of Protein-Protein Binding Specificity within a Family of Synaptic Surface Receptors.

Atomic-level information is essential to explain the formation of specific protein complexes in terms of structure and dynamics. The set of Dpr and DIP proteins, which play a key role in the neuromorphogenesis in the nervous system of Drosophila melanogaster, offer a rich paradigm to learn about protein-protein recognition. Many members of the DIP subfamily cross-react with several members of the Dpr family and vice versa. While there exists a total of 231 possible Dpr-DIP heterodimer complexes from the 21 Dpr and 11 DIP proteins, only 57 "cognate" pairs have been detected by surface plasmon resonance (SPR) experiments, suggesting that the remaining 174 pairs have low or unreliable binding affinity. Our goal is to assess the performance of computational approaches to characterize the global set of interactions between Dpr and DIP proteins and identify the specificity of binding between each DIP with their corresponding Dpr binding partners. In addition, we aim to characterize how mutations influence the specificity of the binding interaction. In this work, a wide range of knowledge-based and physics-based approaches are utilized, including mutual information, linear discriminant analysis, homology modeling, molecular dynamics simulations, Poisson-Boltzmann continuum electrostatics calculations, and alchemical free energy perturbation to decipher the origin of binding specificity of the Dpr-DIP complexes examined. Ultimately, the results show that those two broad strategies are complementary, with different strengths and limitations. Biological inter-relations are more clearly revealed through knowledge-based approaches combining evolutionary and structural features, the molecular determinants controlling binding specificity can be predicted accurately with physics-based approaches based on atomic models.

Hydrological regimes in different slope environments and implications on rainfall thresholds triggering shallow landslides

Assessing hazard of rainfall-induced shallow landslides represents a challenge for the risk management of urbanized areas for which the setting up of early warning systems, based on the reconstruction of reliable rainfall thresholds and rainfall monitoring, is a solution more practicable than the delocalization of settlements and infrastructures. Consequently, the reduction in uncertainties affecting the estimation of rainfall thresholds conditions, leading to the triggering of slope instabilities, is a fundament task to be tackled. In such a view, coupled soil hydrological monitoring and physics-based modeling approaches are presented for estimating rainfall thresholds in two different geomorphological environments prone to shallow landsliding. Based on the comparison of results achieved for silty–clayey soils characterizing Oltrepò Pavese area (northern Italy) and ash-fall pyroclastic soils mantling slopes of Sarno Mountains ridge (southern Italy), this research advances the understanding of the slope hydrological response in triggering shallow landslides. Among the principal results is the comprehension that, mainly depending on geological and geomorphological settings, geotechnical and hydrological properties of soil coverings have a fundamental control on the timing and intensity of hydrological processes leading to landslide initiation. Moreover, results obtained show how the characteristics of the soil coverings control the slope hydrological response at different time scales, making the antecedent soil hydrological conditions a not negligible factor for estimating landslide rainfall thresholds. The approaches proposed can be conceived as an adaptable tool to assess hazard to initiation of shallow rainfall-induced landslides and to implement early-warning systems from site-specific to distributed (catchment or larger) scales.

A hybrid remaining useful life prediction method for cutting tool considering the wear state

Accurate cutting tool remaining useful life (RUL) prediction is of significance to guarantee the cutting quality and minimize the production cost. Recently, physics-based and data-driven methods have been widely used in the tool RUL prediction. The physics-based approaches may not accurately describe the time-varying wear process due to a lack of knowledge for underlying physics and simplifications involved in physical models, while the data-driven methods may be easily affected by the quantity and quality of data. To overcome the drawbacks of these two approaches, a hybrid prognostics framework considering tool wear state is developed to achieve an accurate prediction. Firstly, the mapping relationship between the sensor signal and tool wear is established by support vector regression (SVR). Then, the tool wear statuses are recognized by support vector machine (SVM), and the results are put into a Bayesian framework as prior information. Thirdly, based on the constructed Bayesian framework, the tool wear model parameters are updated iteratively by the sliding time window and particle filter algorithm. Finally, the tool wear state space and RUL can be predicted accordingly using the updating tool wear model. The validity of the proposed method is demonstrated by a high-speed machine tool experiment. The results show that the presented approach can effectively reduce the uncertainty of tool wear state estimation and improve the accuracy of RUL prediction.

Physics-Based Computational Approaches to Compute the Viscoelasticity of Semiflexible Filamentous Biomaterials

This mini-review highlights recent advances on computational approaches that have been used in the characterisation of the viscoelastic response of semiflexible filamentous biomaterials. Special attention is given to the multiscale and coarse-grained approaches that might be used to model the mechanical properties of systems which involve biopolymer assemblies, for instance, actin, collagen, vimentin, microtubules, DNA, viruses, silk, amyloid fibrils, and other protein-based filaments. Besides the basic features of the most commonly used models for semiflexible filaments, I present a brief overview of the numerical approaches that can be used to extract the viscoelasticity of dilute and concentrated solutions, as well as systems with cross-linked networks. Selected examples of simulations that attempt to retrieve the complex shear moduli at experimentally relevant time and length scales, i.e., including not only the fully formed filaments and networks but also their self-assembly kinetics, are also considered.

Energy baseline prediction for buildings: A review

Building energy consumption prediction has gotten a lot of attention in the literature since it is a complicated process influenced by a lot of different factors. Hence, precisely calculating the energy consumption of buildings and enhancing its efficiency is a difficult task. As a result, good energy management and prediction are becoming increasingly vital in the support of energy saving. For this purpose, an energy baseline model is a useful tool for modeling energy consumption for any building over a period of time. It can define as a comparison tool that allows to evaluate energy performance before and after some modification in the system (building). The primary goal of the energy baseline model is to estimate energy demand if energy efficiency measures are applied to a building for the sake of energy savings estimations. The purpose of this study is to examine current approaches for estimating building energy baselines, including physics-based, data-driven, and hybrid approaches, as well as model development process and essential elements. The choice of input independent variables is critical when developing energy baseline models since it affects the estimation of energy savings. Therefore, this review summarizes the input variables that influence energy consumption. Performance metrics are also provided to evaluate the energy baseline models and the effects of independent variables.

Predicting combined tidal and pluvial flood inundation using a machine learning surrogate model

Flooding increases in recent years, in particular for coastal communities facing sea level rise, have brought renewed attention to real-time, street-scale flood forecasting. Such flood models using conventional physics-based modeling approaches are often unrealistic for real-time decision support use cases due to their long model runtime. Machine learning offers an alternative strategy whereby a surrogate model can be trained to mimic relationships present within the physics-based model and, after training, can run in seconds rather than hours. This study used the Random Forest (RF) algorithm to emulate a 1D/2D physics-based model simulating surface water depths in an urban coastal watershed in Norfolk, Virginia. Environmental features from a selected set of pluvial and tidal flood events and topographic information of the roadway were the input variables to train the surrogate model. Results show the potential for the surrogate model to predict flood extent and depth for both pluvial and tidal flood events. Furthermore, the surrogate model can differentiate between flooding locations dominated by pluvial or tidal flooding or impacted by both flooding mechanisms. Flood reports from the mobile app Waze were used for model validation and show 90% agreement with flooding locations from the surrogate model. Finally, feature importance methods were investigated to interpret the performance of the RF models and understand the contribution of different physical features to localized flooding.

Calendar ageing model of Li-ion battery combining physics-based and empirical approaches

A prognosis model named “Dual-tank ageing model”, combining empirical ageing laws for electrode capacity fading and a physics-based model for the Solid Electrolyte Interphase growth, is developed. The ageing laws for the positive and negative electrodes take into account the operating conditions (i.e. temperature and SOC) while a SEI growth model is implemented to account for the lithium consumption at the negative electrode. The model is identified and validated on commercial 43 Ah Graphite/LMO-NMC Li-ion cells in calendar ageing condition at different temperature and SOC. In calendar ageing conditions, the model is able to forecast the ageing of the cell while capturing the evolution of internal parameters such as electrode capacities, electrode potentials and lithium contents.

Engineering empowered by physics-based and data-driven hybrid models: A methodological overview

Smart manufacturing implies creating virtual replicas of the processing operations, taking into account the material dimension and its multi-physics transformation when forming processes operate. Performing efficient, that is, online accurate predictions of the induced properties (including potential defects) of the formed part (to optimally control the process parameters) needs moving beyond usual offline simulation based on nominal models, and proceeds by assimilating data. This will serve, from one side, to keep the model calibrated, and from the other, to enrich the model and its associated predictions, to avoid bias, to improve accuracy or for performing online diagnosis, by advertising on preventive maintenance. For all these purposes, a new alliance between physics-based and data-driven modelling approaches seems a very valuable route for empowering engineering in general, and smart manufacturing in particular. The present paper revisits the main methodologies involved in the construction of the component or system Hybrid Twins.

Predicting performance of in-situ microbial enhanced oil recovery process and screening of suitable microbe-nutrient combination from limited experimental data using physics informed machine learning approach

Screening of suitable microbe-nutrient combination and prediction of oil recovery at the initial stage is essential for the success of Microbial Enhanced Oil Recovery (MEOR) technique. However, experimental and physics-based modelling approaches are expensive and time-consuming. In this study, Physics Informed Machine Learning (PIML) framework was developed to screen and predict oil recovery at a relatively lesser time and cost with limited experimental data. The screening was done by quantifying the influence of parameters on oil recovery from correlation and feature importance studies. Results revealed that microbial kinetic, operational and reservoir parameters influenced the oil recovery by 50%, 32.6% and 17.4%, respectively. Higher oil recovery is attained by selecting a microbe-nutrient combination having a higher ratio of value between biosurfactant yield and microbial yield parameters, as they combinedly influence the oil recovery by 27%. Neural Network is the best ML model for MEOR application to predict oil recovery (R2≈0.99).

A Hybrid Physics-Based and Stochastic Neural Network Model Structure for Diesel Engine Combustion Events

Estimation of combustion phasing and power production is essential to ensuring proper combustion and load control. However, archetypal control-oriented physics-based combustion models can become computationally expensive if highly accurate predictive capabilities are achieved. Artificial neural network (ANN) models, on the other hand, may provide superior predictive and computational capabilities. However, using classical ANNs for model-based prediction and control can be challenging, since their heuristic and deterministic black-box nature may make them intractable or create instabilities. In this paper, a hybridized modeling framework that leverages the advantages of both physics-based and stochastic neural network modeling approaches is utilized to capture CA50 (the timing when 50% of the fuel energy has been released) along with indicated mean effective pressure (IMEP). The performance of the hybridized framework is compared to a classical ANN and a physics-based-only framework in a stochastic environment. To ensure high robustness and low computational burden in the hybrid framework, the CA50 input parameters along with IMEP are captured with a Bayesian regularized ANN (BRANN) and then integrated into an overall physics-based 0D Wiebe model. The outputs of the hybridized CA50 and IMEP models are then successively fine-tuned with BRANN transfer learning models (TLMs). The study shows that in the presence of a Gaussian-distributed model uncertainty, the proposed hybridized model framework can achieve an RMSE of 1.3 × 10−5 CAD and 4.37 kPa with a 45.4 and 3.6 s total model runtime for CA50 and IMEP, respectively, for over 200 steady-state engine operating conditions. As such, this model framework may be a useful tool for real-time combustion control where in-cylinder feedback is limited.

Artificial intelligence-enhanced seismic response prediction of reinforced concrete frames

Existing physics-based modeling approaches do not have a good compromise between performance and computational efficiency in predicting the seismic response of reinforced concrete (RC) frames, where high-fidelity models (e.g., fiber-based modeling method) have reasonable predictive performance but are computationally demanding, while more simplified models (e.g., shear building model) are the opposite. This paper proposes a novel artificial intelligence (AI)-enhanced computational method for seismic response prediction of RC frames which can remedy these problems. The proposed AI-enhanced method incorporates an AI technique with a shear building model, where the AI technique can directly utilize the real-world experimental data of RC columns to determine the lateral stiffness of each column in the target RC frame while the structural stiffness matrix is efficiently formulated via the shear building model. Therefore, this scheme can enhance prediction accuracy due to the use of real-world data while maintaining high computational efficiency due to the incorporation of the shear building model. Two data-driven seismic response solvers are developed to implement the proposed approach based on a database including 272 RC column specimens. Numerical results demonstrate that compared to the experimental data, the proposed method outperforms the fiber-based modeling approach in both prediction capability and computational efficiency and is a promising tool for accurate and efficient seismic response prediction of structural systems.

Boosting the predictive performance with aqueous solubility dataset curation

Intrinsic solubility is a critical property in pharmaceutical industry that impacts in-vivo bioavailability of small molecule drugs. However, solubility prediction with Artificial Intelligence(AI) are facing insufficient data, poor data quality, and no unified measurements for AI and physics-based approaches. We collect 7 aqueous solubility datasets, and present a dataset curation workflow. Evaluating the curated data with two expanded deep learning methods, improved RMSE scores on all curated thermodynamic datasets are observed. We also compare expanded Chemprop enhanced with curated data and state-of-art physics-based approach using pearson and spearman correlation coefficients. A similar performance on pearson with 0.930 and spearman with 0.947 from expanded Chemprop is achieved. A steadily improved pearson and spearman values with increasing data points are also illustrated. Besides that, the computation advantage of AI models enables quick evaluation of a large set of molecules during the hit identification or lead optimization stages, which helps further decision making within the time cycle at drug discovery stage.