Search

Uncertainty is ubiquitous in biological systems. For example, since gene expression is intrinsically governed by noise, nature shows a fascinating degree of variability. If we want to use a model to predict the behaviour of such an intrinsically stochastic system, we have to cope with the fact that the model parameters are never exactly known, but vary according to some distribution. A key question is then to determine how the uncertainties in the parameters affect the model outcome. Knowing the latter uncertainties is crucial when a model is used for, e.g., experimental design, optimisation, or decision-making. To establish how parameter and model prediction uncertainties are related, Monte Carlo approaches could be used. Then, the model is evaluated for a huge number of parameters sets, drawn from the multivariate parameter distribution. However, when model solutions are computationally expensive this approach is intractable. To overcome this problem, so-called spectral expansion (SE) methods have been developed to quantify prediction uncertainty within a probabilistic framework. Such SE methods have a basis in, e.g., computational mathematics, engineering, physics, and fluid dynamics, and, to a lesser extent, systems biology. The computational costs of SE schemes mainly stem from the calculation of the expansion coefficients. Furthermore, SE effectively leads to a surrogate model which captures the dependence of the model on the uncertainty parameters, but is much simpler to execute compared to the original model. In this paper, we present an innovative scheme for the calculation of the expansion coefficients. It guarantees that the model has to be evaluated only a restricted number of times. Especially for models of high complexity this may be a huge computational advantage. By applying the scheme to a variety of examples we show its power, especially in challenging situations where solutions slowly converge due to high computational costs, bifurcations, and discontinuities. [ABSTRACT FROM AUTHOR]

Predicting the key plume evolution features of groundwater contamination are crucial for assessing uncertainty in contamination control and remediation, while implementing detailed complex numerical models for a large number of scenario simulations is time-consuming and sometimes even impossible. This work develops surrogate models with an effective and practicable pathway for predicting the key plume evolution features, such as the distance of maximum plume spreading, of groundwater contamination with natural attenuation. The representative various scenarios of the input parameter combinations were effectively generated by the orthogonal experiment method and the corresponding numerical simulations were performed by the reliable Groundwater Modeling System. The PSO-SVM surrogate models were first developed, and the accuracy was gradually enhanced from 0.5 to 0.9 under a multi-objective condition by effectively increasing the sample data size from 54 sets to 78 sets and decreasing the input variables from 25 of all the considered parameters to a smaller number of the key controlling factors. The statistical surrogate models were also constructed with the fitting degree all above 0.85. The achieved findings provide effective generic surrogate models along with a scientific basis and investigation approach reference for the environmental risk management and remediation of groundwater contamination, particularly with limited data. [ABSTRACT FROM AUTHOR]

The enhanced multi-objective symbolic discretization for time series (eMODiTS) method employs a flexible discretization scheme using different value cuts for each non-equal time interval, which incurs a high computational cost for evaluating each objective function. It is essential to mention that each solution found by eMODiTS is a different-sized vector. Previous work was performed where surrogate models were implemented to reduce the computational cost to solve this problem. However, low-fidelity approximations were obtained concerning the original model. Consequently, our main objective is to propose an improvement to this work, modifying the updating process of the surrogate models to minimize their disadvantages. This improvement was evaluated based on classification, predictive power, and computational cost, comparing it against the original model and ten discretization methods reported in the literature. The results suggest that the proposal achieves a higher fidelity to the original model than previous work. It also achieved a computational cost reduction rate between 15% and 80% concerning the original model. Finally, the classification error of our proposal is similar to eMODiTS and maintains its behavior compared to the other discretization methods. [ABSTRACT FROM AUTHOR]

The densification of urban spaces globally has contributed to a need for design tools supporting the planning of more sustainable, efficient, and liveable cities. Urban Design Optimisation (UDO) responds to this challenge by providing a means to explore many design solutions for a district, evaluate multiple objectives, and make informed selections from many Pareto-efficient solutions. UDO distinguishes itself from other forms of design optimisation by addressing the challenges of incorporating a wide range of planning goals, managing the complex interactions among various urban datasets, and considering the social–technical aspects of urban planning involving multiple stakeholders. Previous reviews focusing on specific topics within UDO do not sufficiently address these challenges. This PRISMA systematic literature review provides an overview of research on topics related to UDO from 2012 to 2022, with articles analysed across seven descriptive categories. This paper presents a discussion on the state-of-the-art and identified gaps present in each of the seven categories. Finally, this paper argues that additional research to improve the socio-technical understanding and usability of UDO would require: (i) methods of optimisation across multiple models, (ii) interfaces that address a multiplicity of stakeholders, (iii) exploration of frameworks for scenario building and backcasting, and (iv) advancing AI applications for UDO, including generalizable surrogates and user preference learning. [ABSTRACT FROM AUTHOR]

This paper explores the alternating conditional expectation (ACE) algorithm-based surrogate model to advance the state-of-practice in uncertainty and sensitivity analysis methodologies for severe accident codes. For engineering purposes, the ACE algorithm has been used as an alternative means to find the optimal functional forms of the multiple input variables and response variables of interest. Analysis results here demonstrate that compared with the reference cases the proposed surrogate model provides much higher performance in terms of the coefficient of determination (R2) and normalized root mean square error (NRMSE), thus giving more robust insights into the relationship and correlation between the input parameters and figures of merit (FOMs) of interest. Relevant results and insights are summarized in terms of points of interest.

Integrated hydrologic models can simulate coupled surface and subsurface processes but are computationally expensive to run at high resolutions over large domains. Here we develop a novel deep learning model to emulate subsurface flows simulated by the integrated ParFlow‐CLM model across the contiguous US. We compare convolutional neural networks like ResNet and UNet run autoregressively against our novel architecture called the Forced SpatioTemporal RNN (FSTR). The FSTR model incorporates separate encoding of initial conditions, static parameters, and meteorological forcings, which are fused in a recurrent loop to produce spatiotemporal predictions of groundwater. We evaluate the model architectures on their ability to reproduce 4D pressure heads, water table depths, and surface soil moisture over the contiguous US at 1 km resolution and daily time steps over the course of a full water year. The FSTR model shows superior performance to the baseline models, producing stable simulations that capture both seasonal and event‐scale dynamics across a wide array of hydroclimatic regimes. The emulators provide over 1,000× speedup compared to the original physical model, which will enable new capabilities like uncertainty quantification and data assimilation for integrated hydrologic modeling that were not previously possible. Our results demonstrate the promise of using specialized deep learning architectures like FSTR for emulating complex process‐based models without sacrificing fidelity. Plain Language Summary: Computational models are important for understanding and predicting terrestrial hydrology, but our most physically detailed models can be time‐consuming and expensive to run over large regions. In this study, we trained deep learning models to emulate a complex hydrology model that simulates groundwater flow over the contiguous US. We developed a new model architecture called FSTR that captures spatiotemporal patterns better than standard deep learning models. FSTR is over 1,000 times faster than ParFlow, the original hydrologic model. This enables new possibilities like forecasting groundwater changes and estimating uncertainties. Our results show that specialized deep learning architectures can accurately emulate complex hydrologic models while drastically reducing computation time. Key Points: Deep learning models can emulate a complex integrated hydrology model that simulates groundwater over the United StatesWe developed a new model architecture that is more robust over longer simulation periods than off‐the‐shelf neural networksDeep‐learning based emulators of complex models enable new applications such as real‐time forecasting and estimating uncertainties [ABSTRACT FROM AUTHOR]

The Ebola virus (EBOV) is a member of the Orthoebolavirus genus, Filoviridae family, which causes severe hemorrhagic diseases in humans and non-human primates (NHPs), with a case fatality rate of up to 90%. The development of countermeasures against EBOV has been hindered by the lack of ideal animal models, as EBOV requires handling in biosafety level (BSL)-4 facilities. Therefore, accessible and convenient animal models are urgently needed to promote prophylactic and therapeutic approaches against EBOV. In this study, a recombinant vesicular stomatitis virus expressing Ebola virus glycoprotein (VSV-EBOV/GP) was constructed and applied as a surrogate virus, establishing a lethal infection in hamsters. Following infection with VSV-EBOV/GP, 3-week-old female Syrian hamsters exhibited disease signs such as weight loss, multi-organ failure, severe uveitis, high viral loads, and developed severe systemic diseases similar to those observed in human EBOV patients. All animals succumbed at 2-3 days post-infection (dpi). Histopathological changes indicated that VSV-EBOV/GP targeted liver cells, suggesting that the tissue tropism of VSV-EBOV/GP was comparable to wild-type EBOV (WT EBOV). Notably, the pathogenicity of the VSV-EBOV/GP was found to be species-specific, age-related, gender-associated, and challenge route-dependent. Subsequently, equine anti-EBOV immunoglobulins and a subunit vaccine were validated using this model. Overall, this surrogate model represents a safe, effective, and economical tool for rapid preclinical evaluation of medical countermeasures against EBOV under BSL-2 conditions, which would accelerate technological advances and breakthroughs in confronting Ebola virus disease. [ABSTRACT FROM AUTHOR]

Uncertainty is ubiquitous in biological systems. For example, since gene expression is intrinsically governed by noise, nature shows a fascinating degree of variability. If we want to use a model to predict the behaviour of such an intrinsically stochastic system, we have to cope with the fact that the model parameters are never exactly known, but vary according to some distribution. A key question is then to determine how the uncertainties in the parameters affect the model outcome. Knowing the latter uncertainties is crucial when a model is used for, e.g., experimental design, optimisation, or decision-making. To establish how parameter and model prediction uncertainties are related, Monte Carlo approaches could be used. Then, the model is evaluated for a huge number of parameters sets, drawn from the multivariate parameter distribution. However, when model solutions are computationally expensive this approach is intractable. To overcome this problem, so-called spectral expansion (SE) methods have been developed to quantify prediction uncertainty within a probabilistic framework. Such SE methods have a basis in, e.g., computational mathematics, engineering, physics, and fluid dynamics, and, to a lesser extent, systems biology. The computational costs of SE schemes mainly stem from the calculation of the expansion coefficients. Furthermore, SE effectively leads to a surrogate model which captures the dependence of the model on the uncertainty parameters, but is much simpler to execute compared to the original model. In this paper, we present an innovative scheme for the calculation of the expansion coefficients. It guarantees that the model has to be evaluated only a restricted number of times. Especially for models of high complexity this may be a huge computational advantage. By applying the scheme to a variety of examples we show its power, especially in challenging situations where solutions slowly converge due to high computational costs, bifurcations, and discontinuities.

We present an approach that uses a deep learning model, in particular, a MultiLayer Perceptron, for estimating the missing values of a variable in multivariate time series data. We focus on filling a long continuous gap (e.g., multiple months of missing daily observations) rather than on individual randomly missing observations. Our proposed gap filling algorithm uses an automated method for determining the optimal MLP model architecture, thus allowing for optimal prediction performance for the given time series. We tested our approach by filling gaps of various lengths (three months to three years) in three environmental datasets with different time series characteristics, namely daily groundwater levels, daily soil moisture, and hourly Net Ecosystem Exchange. We compared the accuracy of the gap-filled values obtained with our approach to the widely used R-based time series gap filling methods ImputeTS and mtsdi. The results indicate that using an MLP for filling a large gap leads to better results, especially when the data behave nonlinearly. Thus, our approach enables the use of datasets that have a large gap in one variable, which is common in many long-term environmental monitoring observations.

Traditionally, nonlinear time history analysis (NLTHA) is used to assess the performance of structures under future hazards which is necessary to develop effective disaster risk management strategies. However, this method is computationally intensive and not suitable for analyzing a large number of structures on a city-wide scale. Surrogate models offer an efficient and reliable alternative and facilitate evaluating the performance of multiple structures under different hazard scenarios. However, creating a comprehensive database for surrogate modelling at the city level presents challenges. To overcome this, the present study proposes meta databases and a general framework for surrogate modelling of steel structures. The dataset includes 30,000 steel moment-resisting frame buildings, representing low-rise, mid-rise and high-rise buildings, with criteria for connections, beams, and columns. Pushover analysis is performed and structural parameters are extracted, and finally, incorporating two different machine learning algorithms, random forest and Shapley additive explanations, sensitivity and explainability analyses of the structural parameters are performed to identify the most significant factors in designing steel moment resisting frames. The framework and databases can be used as a validated source of surrogate modelling of steel frame structures in order for disaster risk management.

This study introduces a deep learning surrogate model designed to predict the evolution of the mean pressure coefficient on the back face of a Windsor body across a range of yaw angles from 2. 5 ∘ to 10 ∘ . Utilizing a variational autoencoder (VAE), the model effectively compresses snapshots of back pressure taken at yaw angles of 2. 5 ∘ , 5 ∘ , and 10 ∘ into two latent vectors. These snapshots are derived from wall-modeled large eddy simulations (WMLESs) conducted at a Reynolds number of R e L = 2.9 × 10 6 . The frequencies that dominate the latent vectors correspond closely with those observed in both the drag's temporal evolution and the dynamic mode decomposition. The projection of the mean pressure coefficient to the latent space yields an increasing linear evolution of the two latent variables with the yaw angle. The mean pressure coefficient distribution at a yaw angle of 7. 5 ∘ is predicted with a mean error of e ¯ = 3.13 % when compared to the WMLESs results after obtaining the values of the latent space with linear interpolation. [ABSTRACT FROM AUTHOR]

Accounting for aeroelastic phenomena, such as flutter, in the conceptual design phase is becoming more important as the trend toward increasing the wing aspect ratio forges ahead. However, this task is computationally expensive, especially when utilizing high-fidelity simulations and numerical optimization. Thus, the development of efficient computational strategies is necessary. With this goal in mind, this work proposes a surrogate-based optimization (SBO) methodology for wing design using a predefined machine learning model. For this purpose, a custom-made Python framework was built based on different open-source codes. The test subject was the classical Goland wing, parameterized to allow for SBO. The process consists of employing a Latin Hypercube Sampling plan and subsequently simulating the resulting wing on SHARPy to generate a dataset. A regression-based machine learning model is then used to build surrogate models for lift and drag coefficients, structural mass, and flutter speed. Finally, after validating the surrogate model, a multi-objective optimization problem aiming to maximize the lift-to-drag ratio and minimize the structural mass is solved through NSGA-II, considering a flutter constraint. This SBO methodology was successfully tested, reaching reductions of three orders of magnitude in the optimization computational time. [ABSTRACT FROM AUTHOR]

In this study, a probabilistic method was proposed for an aircraft's thermal protective layers. The uncertainties of material properties, geometric dimensions, and incoming flow environments were considered for the design inputs. To accelerate the design efficiency, Latin hypercube sampling and surrogate models were built based on finite element method calculations to enhance the simulation efficiency. Thus, the Monte Carlo method can be implemented with such a fast simulation method and produce a massive number of samples for the uncertainty quantification and sensitivity analysis, exploring their impact on the back temperature of the thermal protection layer. Compared to the deterministic method with the extreme deviation design, the probabilistic design yields a weight reduction of 15.61%. This indicates that probabilistic design is an efficient approach to enhance the performance of aircraft and reduce the overall weight of the aircraft. The general goal of this study is to provide a new design method for the coating film of thermal protection systems by considering multiple sources of uncertainties. [ABSTRACT FROM AUTHOR]

Abstract Integrated hydrologic models can simulate coupled surface and subsurface processes but are computationally expensive to run at high resolutions over large domains. Here we develop a novel deep learning model to emulate subsurface flows simulated by the integrated ParFlow‐CLM model across the contiguous US. We compare convolutional neural networks like ResNet and UNet run autoregressively against our novel architecture called the Forced SpatioTemporal RNN (FSTR). The FSTR model incorporates separate encoding of initial conditions, static parameters, and meteorological forcings, which are fused in a recurrent loop to produce spatiotemporal predictions of groundwater. We evaluate the model architectures on their ability to reproduce 4D pressure heads, water table depths, and surface soil moisture over the contiguous US at 1 km resolution and daily time steps over the course of a full water year. The FSTR model shows superior performance to the baseline models, producing stable simulations that capture both seasonal and event‐scale dynamics across a wide array of hydroclimatic regimes. The emulators provide over 1,000× speedup compared to the original physical model, which will enable new capabilities like uncertainty quantification and data assimilation for integrated hydrologic modeling that were not previously possible. Our results demonstrate the promise of using specialized deep learning architectures like FSTR for emulating complex process‐based models without sacrificing fidelity.

Expensive optimization problems are characterized by the significant amount of time and resources needed to determine the quality of potential solutions. This poses severe limitations for the application of metaheuristic optimization methods, such as evolutionary algorithms, as they usually require evaluating many candidate solutions to deliver satisfactory results. Surrogate model-based strategies have become a popular choice to tackle this type of problem. The key idea of these strategies is to build a model which can approximate and (partially) replace the mechanisms for assessing solution quality, such that the use of this less expensive alternative lowers the overall computational cost of the optimization process. This paper analyzes surrogate model-based strategies in the context of a specific, expensive, combinatorial optimization problem: the configuration of the gas distribution system for an electrostatic precipitator. Focusing on this relevant case study from industry, the aim of this paper is twofold: (i) to investigate the most suitable learning techniques for building the surrogate models and (ii) to explore the advantages of ensemble strategies allowing various surrogate models to collaborate during optimization. This contrasts with previous studies where a single, fixed modeling technique is adopted to address this problem. The experimental evaluation involves eight different learning techniques, three alternative ensemble strategies, and two reference approaches from the literature (previously used to tackle this specific problem). Our results reveal the best modeling techniques at the individual level, while highlighting clear benefits of the simultaneous exploitation of multiple surrogate models when facing this particular optimization challenge.

Due to increasing environmental concerns and global energy demand, the development of Floating Offshore Wind Turbines (FOWTs) is on the rise. FOWTs offer a promising solution to expand wind farm deployment into deeper waters with abundant wind resources. However, their harsh operating conditions and lower maturity level compared to fixed structures pose significant engineering challenges, notably in the design phase. A critical challenge is the time-consuming hydromechanics analysis traditionally done using computationally intensive Computational Fluid Dynamics (CFD) models. In this study, we introduce Artificial Intelligence-based surrogate models using state-of-the-art Machine Learning algorithms. These surrogate models achieve CFD-level accuracy (within 3% difference) while dramatically reducing computational requirements from minutes to milliseconds. Specifically, we build a surrogate model for characterizing the hydrodynamic response of a floating spar-type offshore wind turbine (including added mass, radiation damping matrices, and hydrodynamic excitation) using computationally efficient shallow Machine Learning models, optimizing the trade-off between computational efficiency and accuracy, based on data generated by a cutting-edge potential-flow code.

Predicting the key plume evolution features of groundwater contamination are crucial for assessing uncertainty in contamination control and remediation, while implementing detailed complex numerical models for a large number of scenario simulations is time-consuming and sometimes even impossible. This work develops surrogate models with an effective and practicable pathway for predicting the key plume evolution features, such as the distance of maximum plume spreading, of groundwater contamination with natural attenuation. The representative various scenarios of the input parameter combinations were effectively generated by the orthogonal experiment method and the corresponding numerical simulations were performed by the reliable Groundwater Modeling System. The PSO-SVM surrogate models were first developed, and the accuracy was gradually enhanced from 0.5 to 0.9 under a multi-objective condition by effectively increasing the sample data size from 54 sets to 78 sets and decreasing the input variables from 25 of all the considered parameters to a smaller number of the key controlling factors. The statistical surrogate models were also constructed with the fitting degree all above 0.85. The achieved findings provide effective generic surrogate models along with a scientific basis and investigation approach reference for the environmental risk management and remediation of groundwater contamination, particularly with limited data.

The enhanced multi-objective symbolic discretization for time series (eMODiTS) method employs a flexible discretization scheme using different value cuts for each non-equal time interval, which incurs a high computational cost for evaluating each objective function. It is essential to mention that each solution found by eMODiTS is a different-sized vector. Previous work was performed where surrogate models were implemented to reduce the computational cost to solve this problem. However, low-fidelity approximations were obtained concerning the original model. Consequently, our main objective is to propose an improvement to this work, modifying the updating process of the surrogate models to minimize their disadvantages. This improvement was evaluated based on classification, predictive power, and computational cost, comparing it against the original model and ten discretization methods reported in the literature. The results suggest that the proposal achieves a higher fidelity to the original model than previous work. It also achieved a computational cost reduction rate between 15% and 80% concerning the original model. Finally, the classification error of our proposal is similar to eMODiTS and maintains its behavior compared to the other discretization methods.

In this work, the authors present a digital twin (DT) to analyse the manufacturing process of structural composites by using resin transfer moulding (RTM). During RTM, a dry textile preform is impregnated with a polymer resin injected in a closed mould. RTM is one of the most used production methods for high-performance structural composites. The DT is focused on detecting in-homogeneous resin flow produced by race-tracking channels that divert resin flow to the outlet gates of the mould, producing dry spots and lack of impregnation. The DT core contains two surrogate models based on encoder/decoder deep learning architectures, providing the fast/accurate response necessary for interrogation during manufacturing. The first surrogate acts as the disturbance detector, providing the on-the-fly representation of the fabric permeability with the only information gathered by a set of five pressure sensors distributed over the mould surface. The second offers real-time representation of a set of quantities of interests (QoI): namely, the flow progress and the pressure field inside the mould. Training of both surrogates was performed with synthetic data generated by high-fidelity multi-physics simulations of the flow progress in a porous preform by following Darcy's law. Errors in the pressure field predictions of the surrogates are lower than 1 % with consultation time <50 ms, enabling encapsulation in the digital twin. The DT performance was evaluated by comparing the response against a set of RTM experiments for different race-tracking scenarios. The two most relevant novelties of the work are the use of the concept of the instantiated DT, which provides information on the current state of the process from data provided by a network of distributed sensors, and that this DT has been trained exclusively with synthetic data from multiphysics simulations while evaluated against experimental data from injection tests. [ABSTRACT FROM AUTHOR]

Featured Application: This study provides data and rationales to aid researchers in the selection of surrogate models for in vitro assessments of surgically repaired meniscal roots. The results are potentially applicable in experimental in vitro investigations aimed at evaluating the performance of surgical repair both in existing approaches and emerging techniques. It could also be of interest for studies seeking to adjust material models of the meniscal tissue around the suture area for incorporation into computational models. Meniscal root repair is not routinely recommended for patients over 75 years old, yet surrogate age-unrestricted human or porcine models are used for its evaluation. This study assesses the suitability of older human or porcine meniscus models for in vitro testing of the sutured meniscal horn. Three groups of menisci underwent a load-to-failure test with continuous monitoring of the traction force and deformation around the suture: human < 75 years, human ≥ 75 years, and porcine. Both surrogate models were compared to the younger group. The porcine group exhibited a 172.1%-higher traction force before tearing (p < 0.001) and a 174.1%-higher ultimate force (p < 0.001), without there being differences between the human groups. At tissue level, the older group had a 28.7%-lower cut-out stress (p = 0.012) and the porcine group had a 57.2%-higher stress (p < 0.001). Regarding elasticity at the sutured area, a 48.1%-greater deformation rate was observed in the older group (p < 0.001), without difference for the porcine group. In conclusion, neither the porcine nor the older human model demonstrated a clear advantage as a surrogate model for young human sutured meniscal horns. The older human meniscus is preferable for resistance at the specimen level, while the porcine model better represents deformation in the sutured zone. [ABSTRACT FROM AUTHOR]

A novel approach is presented for predicting compressible turbulent flow fields using a neural network-based data-driven method. Accurate prediction in turbulent regions heavily relies on the resolution of available data. Traditional methods, employing image-based techniques by mapping scattered computational fluid dynamics (CFD) data onto Cartesian grids, encounter data scarcity in critical areas such as the boundary layer and wake. Recently, convolutional neural networks (CNN) have gained prominence as the most widely referenced technique in fluid dynamics, utilizing flow field images as datasets for flow field prediction. However, CNN requires datasets with a high pixel density to enhance training accuracy in crucial regions, thereby increasing the input data volume and machine training time. To address this challenge, our proposed method deviates from using flow field images and instead generates datasets directly from the flow field properties of CFD grid points. By employing this approach, several advantages are realized. Firstly, the network benefits from the favorable characteristics of unstructured grids, such as varying point spacing near the object surface and in the far field, which effectively reduces the amount of input data and consequently the machine training cost. Secondly, the construction of the training dataset eliminates the need for interpolation or extrapolation, thereby preserving the accuracy of CFD data. In this case, a simple multilayer perceptron can be trained using the proposed dataset. Various flow field properties, including static pressure, turbulent kinetic energy, and velocity components, can be predicted with high accuracy within a few seconds. [ABSTRACT FROM AUTHOR]

A novel approach is presented for predicting compressible turbulent flow fields using a neural network-based data-driven method. Accurate prediction in turbulent regions heavily relies on the resolution of available data. Traditional methods, employing image-based techniques by mapping scattered computational fluid dynamics (CFD) data onto Cartesian grids, encounter data scarcity in critical areas such as the boundary layer and wake. Recently, convolutional neural networks (CNN) have gained prominence as the most widely referenced technique in fluid dynamics, utilizing flow field images as datasets for flow field prediction. However, CNN requires datasets with a high pixel density to enhance training accuracy in crucial regions, thereby increasing the input data volume and machine training time. To address this challenge, our proposed method deviates from using flow field images and instead generates datasets directly from the flow field properties of CFD grid points. By employing this approach, several advantages are realized. Firstly, the network benefits from the favorable characteristics of unstructured grids, such as varying point spacing near the object surface and in the far field, which effectively reduces the amount of input data and consequently the machine training cost. Secondly, the construction of the training dataset eliminates the need for interpolation or extrapolation, thereby preserving the accuracy of CFD data. In this case, a simple multilayer perceptron can be trained using the proposed dataset. Various flow field properties, including static pressure, turbulent kinetic energy, and velocity components, can be predicted with high accuracy within a few seconds.

Hybrid variational quantum algorithms, which combine a classical optimizer with evaluations on a quantum chip, are the most promising candidates to show quantum advantage on current noisy, intermediate-scale quantum (NISQ) devices. The classical optimizer is required to perform well in the presence of noise in the objective function evaluations, or else it becomes the weakest link in the algorithm. We introduce the use of Gaussian Processes (GP) as surrogate models to reduce the impact of noise and to provide high quality seeds to escape local minima, whether real or noise-induced. We build this as a framework on top of local optimizations, for which we choose Implicit Filtering (ImFil) in this study. ImFil is a state-of-the-art, gradient-free method, which in comparative studies has been shown to outperform on noisy VQE problems. The result is a new method: "GP+ImFil". We show that when noise is present, the GP+ImFil approach finds results closer to the true global minimum in fewer evaluations than standalone ImFil, and that it works particularly well for larger dimensional problems. Using GP to seed local searches in a multi-modal landscape shows mixed results: although it is capable of improving on ImFil standalone, it does not do so consistently and would only be preferred over other, more exhaustive, multistart methods if resources are constrained.

The densification of urban spaces globally has contributed to a need for design tools supporting the planning of more sustainable, efficient, and liveable cities. Urban Design Optimisation (UDO) responds to this challenge by providing a means to explore many design solutions for a district, evaluate multiple objectives, and make informed selections from many Pareto-efficient solutions. UDO distinguishes itself from other forms of design optimisation by addressing the challenges of incorporating a wide range of planning goals, managing the complex interactions among various urban datasets, and considering the social–technical aspects of urban planning involving multiple stakeholders. Previous reviews focusing on specific topics within UDO do not sufficiently address these challenges. This PRISMA systematic literature review provides an overview of research on topics related to UDO from 2012 to 2022, with articles analysed across seven descriptive categories. This paper presents a discussion on the state-of-the-art and identified gaps present in each of the seven categories. Finally, this paper argues that additional research to improve the socio-technical understanding and usability of UDO would require: (i) methods of optimisation across multiple models, (ii) interfaces that address a multiplicity of stakeholders, (iii) exploration of frameworks for scenario building and backcasting, and (iv) advancing AI applications for UDO, including generalizable surrogates and user preference learning.

Sustainable management of groundwater resources under changing climatic conditions require an application of reliable and accurate predictions of groundwater levels. Mechanistic multi-scale, multi-physics simulation models are often too hard to use for this purpose, especially for groundwater managers who do not have access to the complex compute resources and data. Therefore, we analyzed the applicability and performance of four modern deep learning computational models for predictions of groundwater levels. We compare three methods for optimizing the models’ hyperparameters, including two surrogate model-based algorithms and a random sampling method. The models were tested using predictions of the groundwater level in Butte County, California, USA, taking into account the temporal variability of streamflow, precipitation, and ambient temperature. Our numerical study shows that the optimization of the hyperparameters can lead to reasonably accurate performance of all models (root mean squared errors of groundwater predictions of 2 meters or less), but the “simplest” network, namely a multilayer perceptron (MLP) performs overall better for learning and predicting groundwater data than the more advanced long short-term memory or convolutional neural networks in terms of prediction accuracy and time-to-solution, making the MLP a suitable candidate for groundwater prediction.

The purpose of this study is to investigate the uncertainty of the design variables of a front suspension lower control arm under fatigue-loading circumstances to estimate a reliable and robust product. This study offers a method for systematic uncertainty quantification (UQ), and the following steps were taken to achieve this: First, a finite element model was built to predict the fatigue life of the control arm under bump-loading conditions. Second, a sensitivity scheme, based on one of the global analyses, was developed to identify the model’s most and least significant design input variables. Third, physics-based and data-driven uncertainty quantification schemes were employed to quantify the model’s input parameter uncertainties via a Monte Carlo simulation. The simulations were conducted using 10,000 samples of material properties and geometrical uncertainty variables, with the coefficients of variation ranging from 1 to 3%. Finally, the confidence interval results show a deviation of about 21.74% from the mean (the baseline). As a result, by applying systematic UQ, a more reliable and robust automobile suspension control arm can be designed during the early stages of design to produce a more efficient and better approximation of fatigue life under uncertain conditions.

Some optimization problems are difficult to solve due to a considerable number of local optima, which may result in premature convergence of the optimization process. To address this problem, we propose a novel heuristic method for constructing a smooth surrogate model of the original function. The surrogate function is easier to optimize but maintains a fundamental property of the original rugged fitness landscape: the location of the global optimum. To create such a surrogate model, we consider a linear genetic programming approach coupled with a self-tuning fitness function. More specifically, to evaluate the fitness of the produced surrogate functions, we employ Fuzzy Self-Tuning Particle Swarm Optimization, a setting-free version of particle swarm optimization. To assess the performance of the proposed method, we considered a set of benchmark functions characterized by high noise and ruggedness. Moreover, the method is evaluated over different problems’ dimensionalities. The proposed approach reveals its suitability for performing the proposed task. In particular, experimental results confirm its capability to find the global argminimum for all the considered benchmark problems and all the domain dimensions taken into account, thus providing an innovative and promising strategy for dealing with challenging optimization problems. Doi: 10.28991/ESJ-2023-07-04-01 Full Text: PDF

The purpose of this study is to investigate the uncertainty of the design variables of a front suspension lower control arm under fatigue-loading circumstances to estimate a reliable and robust product. This study offers a method for systematic uncertainty quantification (UQ), and the following steps were taken to achieve this: First, a finite element model was built to predict the fatigue life of the control arm under bump-loading conditions. Second, a sensitivity scheme, based on one of the global analyses, was developed to identify the model's most and least significant design input variables. Third, physics-based and data-driven uncertainty quantification schemes were employed to quantify the model's input parameter uncertainties via a Monte Carlo simulation. The simulations were conducted using 10,000 samples of material properties and geometrical uncertainty variables, with the coefficients of variation ranging from 1 to 3%. Finally, the confidence interval results show a deviation of about 21.74% from the mean (the baseline). As a result, by applying systematic UQ, a more reliable and robust automobile suspension control arm can be designed during the early stages of design to produce a more efficient and better approximation of fatigue life under uncertain conditions. [ABSTRACT FROM AUTHOR]

The paper introduces a novel framework for detecting adversarial attacks on machine learning models that classify tabular data. Its purpose is to provide a robust method for the monitoring and continuous auditing of machine learning models for the purpose of detecting malicious data alterations. The core of the framework is based on building machine learning classifiers for the detection of attacks and its type that operate on diagnostic attributes. These diagnostic attributes are obtained not from the original model, but from the surrogate model that has been created by observation of the original model inputs and outputs. The paper presents building blocks for the framework and tests its power for the detection and isolation of attacks in selected scenarios utilizing known attacks and public machine learning data sets. The obtained results pave the road for further experiments and the goal of developing classifiers that can be integrated into real-world scenarios, bolstering the robustness of machine learning applications. [ABSTRACT FROM AUTHOR]

This study introduces a deep learning surrogate model designed to predict the evolution of the mean pressure coefficient on the back face of a Windsor body across a range of yaw angles from 2.5∘ to 10∘. Utilizing a variational autoencoder (VAE), the model effectively compresses snapshots of back pressure taken at yaw angles of 2.5∘, 5∘, and 10∘ into two latent vectors. These snapshots are derived from wall-modeled large eddy simulations (WMLESs) conducted at a Reynolds number of ReL=2.9×106. The frequencies that dominate the latent vectors correspond closely with those observed in both the drag’s temporal evolution and the dynamic mode decomposition. The projection of the mean pressure coefficient to the latent space yields an increasing linear evolution of the two latent variables with the yaw angle. The mean pressure coefficient distribution at a yaw angle of 7.5∘ is predicted with a mean error of e¯=3.13% when compared to the WMLESs results after obtaining the values of the latent space with linear interpolation.

Accounting for aeroelastic phenomena, such as flutter, in the conceptual design phase is becoming more important as the trend toward increasing the wing aspect ratio forges ahead. However, this task is computationally expensive, especially when utilizing high-fidelity simulations and numerical optimization. Thus, the development of efficient computational strategies is necessary. With this goal in mind, this work proposes a surrogate-based optimization (SBO) methodology for wing design using a predefined machine learning model. For this purpose, a custom-made Python framework was built based on different open-source codes. The test subject was the classical Goland wing, parameterized to allow for SBO. The process consists of employing a Latin Hypercube Sampling plan and subsequently simulating the resulting wing on SHARPy to generate a dataset. A regression-based machine learning model is then used to build surrogate models for lift and drag coefficients, structural mass, and flutter speed. Finally, after validating the surrogate model, a multi-objective optimization problem aiming to maximize the lift-to-drag ratio and minimize the structural mass is solved through NSGA-II, considering a flutter constraint. This SBO methodology was successfully tested, reaching reductions of three orders of magnitude in the optimization computational time.

Reinforced concrete (RC) is the result of a combination of steel reinforcing rods (which have high tensile) and concrete (which has high compressive strength). Additionally, the prediction of long-term deformations of RC flexural structures and the magnitude of the influence of the relevant material and geometric parameters are important for evaluating their serviceability and safety throughout their life cycles. Empirical methods for predicting the long-term deformation of RC structures are limited due to the difficulty of considering all the influencing factors. In this study, four popular surrogate models, i.e., polynomial chaos expansion (PCE), support vector regression (SVR), Kriging, and radial basis function (RBF), are used to predict the long-term deformation of RC structures. The surrogate models were developed and evaluated using RC simply supported beam examples, and experimental datasets were collected for comparison with common machine learning models (back propagation neural network (BP), multilayer perceptron (MLP), decision tree (DT) and linear regression (LR)). The models were tested using the statistical metrics R 2 , R A A E , R M A E , R M S E ,   V A F ,   P I ,   A 10 − i n d e x   a n d   U 95 . The results show that all four proposed models can effectively predict the deformation of RC structures, with PCE and SVR having the best accuracy, followed by the Kriging model and RBF. Moreover, the prediction accuracy of the surrogate model is much lower than that of the empirical method and the machine learning model in terms of the RMSE. Furthermore, a global sensitivity analysis of the material and geometric parameters affecting structural deflection using PCE is proposed. It was found that the geometric parameters are more influential than the material parameters. Additionally, there is a coupling effect between material and geometric parameters that works together to influence the long-term deflection of RC structures. [ABSTRACT FROM AUTHOR]

The processing of unstructured data, such as natural language texts, is one of the urgent tasks in the development of intelligent products. In turn, topic modeling as a method of working with unmarked and partially marked text data is a natural choice for analyzing document bodies and creating vector representations. In this regard, it is especially important to train high-quality thematic models in a short time which is possible with the help of the proposed framework. The developed framework implements an evolutionary approach to optimizing hyperparameters of models with additive regularization and high results on quality metrics (coherence, NPMI). To reduce the computational time, a mode of working with surrogate models is presented which provides acceleration of calculations up to 1.8 times without loss of quality. The effectiveness of the framework is demonstrated on three datasets with different statistical characteristics. The results obtained exceed similar solutions by an average of 20 % in coherence and 5 % in classification quality for two of the three datasets. A distributed version of the framework has been developed for conducting experimental studies of topic models. The developed framework can be used by users without special knowledge in the field of topic modeling due to the default data processing pipeline. The results of the work can be used by researchers to analyze topic models and expand functionality.

Abstract Graph Convolutional Neural Network (GCNN) is a popular class of deep learning (DL) models in material science to predict material properties from the graph representation of molecular structures. Training an accurate and comprehensive GCNN surrogate for molecular design requires large-scale graph datasets and is usually a time-consuming process. Recent advances in GPUs and distributed computing open a path to reduce the computational cost for GCNN training effectively. However, efficient utilization of high performance computing (HPC) resources for training requires simultaneously optimizing large-scale data management and scalable stochastic batched optimization techniques. In this work, we focus on building GCNN models on HPC systems to predict material properties of millions of molecules. We use HydraGNN, our in-house library for large-scale GCNN training, leveraging distributed data parallelism in PyTorch. We use ADIOS, a high-performance data management framework for efficient storage and reading of large molecular graph data. We perform parallel training on two open-source large-scale graph datasets to build a GCNN predictor for an important quantum property known as the HOMO-LUMO gap. We measure the scalability, accuracy, and convergence of our approach on two DOE supercomputers: the Summit supercomputer at the Oak Ridge Leadership Computing Facility (OLCF) and the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC). We present our experimental results with HydraGNN showing (i) reduction of data loading time up to 4.2 times compared with a conventional method and (ii) linear scaling performance for training up to 1024 GPUs on both Summit and Perlmutter.

Abstract To increase the efficiency and robustness of stability-based transition prediction in flow simulations, simplified methods are introduced to substitute direct stability analyses for rapid disturbance growth prediction. For low-speed boundary layers, these methods are mainly established based on self-similar assumptions, which are not applicable to non-similar boundary layers in hypersonic flows. The objective of this article is to investigate the application of surrogate models to stability analysis of non-similar flows over blunt cones, focused on parameterization of boundary-layer (BL) profiles. Firstly, correlations between BL edge and profile parameters are analyzed, along with self-similar flow parameters and discrete points on BL profiles, which present four groups of BL characteristic parameters. Secondly, using these parameters as inputs, surrogate models are built for disturbance growth prediction over an MF-1 blunt cone. Results show that, surrogate models using four BL edge parameters and a BL shape factor {U e , T e , ρ e , η e , H 12} for stability analysis can achieve comparable accuracy with those using 16 discrete BL profile parameters, which are more precise than those using merely self-similar parameters or BL edge parameters. Thirdly, the established surrogate models are validated by stability analysis and transition prediction over the MF-1 blunt cone in flight experiments at the instants of t = 17 s ~ 22 s. Compared with direct linear stability analyses, the mean relative error of predicted disturbance growth rates by surrogate models is 8.0% and the maximum relative error of N factor envelopes is 6.6%, which indicates feasible applications of surrogate models to stability analysis and transition prediction of non-similar boundary layers in hypersonic flows.

Some optimization problems in scientific research, such as the robustness optimization for the Internet of Things and the neural architecture search, are large-scale in decision space and expensive for objective evaluation. In order to get a good solution in a limited budget for the large-scale expensive optimization, a random grouping strategy is adopted to divide the problem into some low-dimensional sub-problems. A surrogate model is then trained for each sub-problem using different strategies to select training data adaptively. After that, a dynamic infill criterion is proposed corresponding to the models currently used in the surrogate-assisted sub-problem optimization. Furthermore, an escape mechanism is proposed to keep the diversity of the population. The performance of the method is evaluated on CEC’2013 benchmark functions. Experimental results show that the algorithm has better performance in solving expensive large-scale optimization problems.

Nowadays, the large set of available simulation tools brings numerous benefits to urban and architectural practices. However, simulations often take a considerable amount of time to yield significant results, particularly when performing many simulations and with large models, as is typical in complex urban and architectural endeavors. Additionally, multiple objective optimizations with metaheuristic algorithms have been widely used to solve building optimization problems. However, most of these optimization processes exponentially increase the computational time to correctly produce outputs and require extensive knowledge to interpret results. Thus, building optimization with time-consuming simulation tools is often rendered unfeasible and requires a specific methodology to overcome these barriers. This work integrates a baseline multi-objective optimization process with a widely used, validated building energy simulation tool. The goal is to minimize the energy use and cost of the construction of a residential building complex. Afterward, machine learning and optimization techniques are used to create a surrogate model capable of accurately predicting the simulation results. Finally, different metaheuristics with their tuned hyperparameters are compared. Results show significant improvements in optimization results with a decrease of up to 22% in the total cost while having similar performance results and execution times up to 100 times faster. [ABSTRACT FROM AUTHOR]

We propose a web-based platform of integrated intelligent agents that incorporates multiple Functional Mock-up Units (FMUs) and surrogate modelling techniques. In this platform, each FMU envelops a stand-alone simulation component that represents an aquaculture system, such as the fish growth model, water quality model, and fish behavior model. Some FMUs may be computationally expensive to simulate or have different time step intervals, making integration with other FMUs difficult. To address these challenges, we employed surrogate models to substitute the more computationally expensive models. In this work, surrogate models are trained using simulation data and selected based on robustness analysis to ensure the overall system input-output reliability. The platform also includes a Chatbot component utilizing natural language processing and decision-making techniques to interpret user requests and provide tailored FMU configurations, enhancing the user simulation experience. Overall, the proposed platform provides a comprehensive and Efficient approach to modelling and simulating complex systems such as fish farms. Robustness analysis ensures the platform's accuracy and reliability, while the user-friendly interface enables easy tailored experimentation. By providing a framework for exploring the potential of aquaculture as a key source of food and income, the proposed platform represents a valuable interactive simulation tool for researchers, policymakers, and industry professionals seeking to improve the sustainability, efficiency, and economic viability of the aquaculture industry. [ABSTRACT FROM AUTHOR]

Numerical modeling is increasingly used to analyze practical rock engineering problems. The geological strength index (GSI) is a critical input for many rock engineering problems. However, no available method allows the quantification of GSI input parameters, and engineers must consider a range of values. As projects progress, these ranges can be narrowed down. Machine learning (ML) algorithms have been coupled with numerical modeling to create surrogate models. The concept of surrogate models aligns well with the deductive nature of data availability in rock engineering projects. In this paper, we demonstrated the use of surrogate models to analyze two common rock slope stability problems: (1) determining the maximum stable depth of a vertical excavation and (2) determining the allowable angle of a slope with a fixed height. Compared with support vector machines and K-nearest algorithms, the random forest model performs best on a data set of 800 numerical models for the problems discussed in the paper. For all these models, regression-type models outperform classification models. Once the surrogate model is confirmed to preform accurately, instantaneous predictions of maximum excavation depth and slope angle can be achieved according to any range of input parameters. This capability is used to investigate the impact of narrowing GSI range estimation. [ABSTRACT FROM AUTHOR]

This research compares the performance of space-time surrogate models (STSMs) and network surrogate models (NSMs). Specifically, when the system response varies over time (or pseudo-time), the surrogates must predict the system response. A surrogate model is used to approximate the response of computationally expensive spatial and temporal fields resulting from some computational mechanics simulations. Within a design context, a surrogate takes a vector of design variables that describe a current design and returns an approximation of the design's response through a pseudo-time variable. To compare various radial basis function (RBF) surrogate modeling approaches, the prediction of a load displacement path of a snap-through structure is used as an example numerical problem. This work specifically considers the scenario where analytical sensitivities are available directly from the computational mechanics' solver and therefore gradient enhanced surrogates are constructed. In addition, the gradients are used to perform a domain transformation preprocessing step to construct surrogate models in a more isotropic domain, which is conducive to RBFs. This work demonstrates that although the gradient-based domain transformation scheme offers a significant improvement to the performance of the space-time surrogate models (STSMs), the network surrogate model (NSM) is far more robust. This research offers explanations for the improved performance of NSMs over STSMs and recommends future research to improve the performance of STSMs. [ABSTRACT FROM AUTHOR]

Crude Oil Distillation Unit (CDU) is one of the most important separation installations in the petroleum refinery industries. In this work, a Bi-level Surrogate column model Aided Constrained Optimization Design (Bi-SACOD) is proposed for time-consuming objectives and constraints in the evolutionary optimization design of CDUs. The main components of Bi-SACOD include bi-level surrogate model construction (Bi-SMC), bi-level model management (Bi-MM), and particle swarm optimization (PSO) mixed-integer constrained evolutionary (PSO-MICE) search. Bi-SMC implements surrogate column model construction and feasible domain identification. Bi-MM combines surrogate column models with rigorous CDU simulations to perform model management, and PSO-MICE implements optimum search works. The optimization results of the CDUs indicate that Bi-SACOD outperforms the single-level surrogate column model approaches, and are more consistent with the rigorous CDU model optimization approach, whereas the evaluation numbers of the timeconsuming rigorous models are significantly reduced. [ABSTRACT FROM AUTHOR]

Since their development at the end of the 50s, panel methods were widely used for the fast simulation of aerospace objects reentry. Although improvements were proposed for the continuum regime formulations, the bridging functions usually employed in the transitional regime did not go through major changes since then. With the current interest in designing Very Low Earth Orbit satellites and more efficient reentry vehicles, a greater level of preciseness is now required for the fast computation of the aerodynamic and aerothermodynamic wall quantities in rarefied regime. In this context, this paper presents a new approach to build Machine Learning based surrogates going from the choice of the design variables and the Design of Experiments, to the models training and evaluation. Hence, kriging and Artificial Neural Networks are respectively trained to predict the pressure and heat flux stagnation coefficients, and the pressure, friction and heat flux coefficient distributions in the rarefied portion of any aerodynamic shape's reentry. • New reduced models for the wall quantities in hypersonic rarefied conditions. • Kriging models for the stagnation heat flux and pressure coefficients. • Neural networks for the friction, heat flux and pressure coefficient distributions. • A thorough and challenging validation of the models has been carried out. • A more precise and generic alternative to classic panel methods. [ABSTRACT FROM AUTHOR]

This paper presents a combined method based on optimized neural networks and optimization algorithms to solve structural optimization problems. The main idea is to utilize an optimized artificial neural network (OANN) as a surrogate model to reduce the number of computations for structural analysis. First, the OANN is trained appropriately. Subsequently, the main optimization problem is solved using the OANN and a population-based algorithm. The algorithms considered in this step are the arithmetic optimization algorithm (AOA) and genetic algorithm (GA). Finally, the abovementioned problem is solved using the optimal point obtained from the previous step and the pattern search (PS) algorithm. To evaluate the performance of the proposed method, two numerical examples are considered. In the first example, the performance of two algorithms, OANN + AOA + PS and OANN + GA + PS, is investigated. Using the GA reduces the elapsed time by approximately 50% compared with using the AOA. Results show that both the OANN + GA + PS and OANN + AOA + PS algorithms perform well in solving structural optimization problems and achieve the same optimal design. However, the OANN + GA + PS algorithm requires significantly fewer function evaluations to achieve the same accuracy as the OANN + AOA + PS algorithm. [ABSTRACT FROM AUTHOR]

In the last few years, energy efficiency has become a challenge. Not only mitigating environmental impact but reducing energy waste can lead to financial advantages. Buildings play an important role in this: they are among the biggest consumers. So, finding manners to reduce energy consumption is a way to minimise energy waste, and a technique for that is creating Demand Response (DR) strategies. This paper proposes a novel way to decrease computational effort of simulating the behaviour of a building using surrogate models based on active learning. Before going straight to the problem of a building, which is complex and computationally costly, the paper proposes the approach of active learning to a smaller problem: with reduced simulations, regress the curve of voltage versus current of a thermo-resistor. Then, the paper implements a surrogate model of energy consumption of a building. The goal is to be able to learn the consumption pattern based on a limited number of simulations. The result given by the surrogate can be used to set the reference temperature, maximising the PV self-consumption, and reducing energy usage from the grid. Thanks to the surrogate, the total time spent to map all possible consumption scenarios is reduced around 7 times.