Non-intrusive Approaches Research Articles

Investigating and exploiting the impact of variability in resonator parameters on the vibration attenuation in locally resonant metamaterials.

Locally resonant metamaterials (LRMs) have recently emerged in the search for lightweight noise and vibration solutions. These materials have the ability to create stop bands, which arise from the sub-wavelength addition of identical resonators to a host structure and result in strong vibration attenuation. However, their manufacturing inevitably introduces variability such that the system as-manufactured often deviates significantly from the original as-designed. This can reduce attenuation performance, but may also broaden the attenuation band. This work focuses on the impact of variability within tolerance ranges in resonator properties on the vibration attenuation in metamaterial beams. Following a qualitative pre-study, two non-intrusive uncertainty propagation approaches are applied to find the upper and lower bounds of three performance metrics, by evaluating deterministic metamaterial models with uncertain parameters defined as interval variables. A global search approach is used and compared with a machine learning (ML)-based uncertainty propagation approach which significantly reduces the required number of simulations. Variability in resonator stiffnesses and masses is found to have the highest impact. Variability in the resonator positions only has a comparable impact for less deep sub-wavelength designs. The broadening potential of varying resonator properties is exploited in broadband optimization and the robustness of the optimized metamaterial is assessed.This article is part of the theme issue 'Current developments in elastic and acoustic metamaterials science (Part 2)'.

A stochastic Galerkin lattice Boltzmann method for incompressible fluid flows with uncertainties

Efficient modeling and simulation of uncertainties in computational fluid dynamics (CFD) remains a crucial challenge. In this paper, we present the first stochastic Galerkin (SG) lattice Boltzmann method (LBM) built upon the generalized polynomial chaos (gPC). The proposed method offers an efficient and accurate approach that depicts the propagation of uncertainties in stochastic incompressible flows. Formal analysis shows that the SG LBM preserves the correct Chapman–Enskog asymptotics and recovers the corresponding macroscopic fluid equations. Numerical experiments, including the Taylor–Green vortex flow, lid-driven cavity flow, and isentropic vortex convection, are presented to validate the solution algorithm. The results demonstrate that the SG LBM achieves the expected spectral convergence and the computational cost is significantly reduced compared to the sampling-based non-intrusive approaches, e.g., the routinely used Monte Carlo method. We obtain a speedup factor of 5.72 compared to Monte Carlo sampling in a randomized two-dimensional Taylor–Green vortex flow test case. By leveraging the accuracy and flexibility of LBM and the efficiency of gPC-based SG, the proposed SG LBM provides a powerful framework for uncertainty quantification in CFD practice.

Virtual design hackathons: a data collection framework

AbstractDesign hackathons offer a unique research opportunity to study time-pressured collaborative design. At the same time, research on design hackathons faces unique methodological challenges, prompting the exploration of new research approaches. This paper proposes a new data-collection framework that leverages a virtual format of hackathon events and enables a deeper insight into hackathon dynamics. The framework applicability is presented through a case study of the IDEA challenge hackathon, in which different intrusive and non-intrusive data collection approaches were used.

A non-intrusive bi-fidelity reduced basis method for time-independent problems

Scientific and engineering problems often involve parametric partial differential equations (PDEs), such as uncertainty quantification, optimizations, and inverse problems. However, solving these PDEs repeatedly can be prohibitively expensive, especially for large-scale complex applications. To address this issue, reduced order modeling (ROM) has emerged as an effective method to reduce computational costs. However, ROM often requires significant modifications to the existing code, which can be time-consuming and complex, particularly for large-scale legacy codes. Non-intrusive methods have gained attention as an alternative approach. However, most existing non-intrusive approaches are purely data-driven and may not respect the underlying physics laws during the online stage, resulting in less accurate approximations of the reduced solution. In this study, we propose a new non-intrusive bi-fidelity reduced basis method for time-independent parametric PDEs. Our algorithm utilizes the discrete operator, solutions, and right-hand sides obtained from the high-fidelity legacy solver. By leveraging a low-fidelity model, we efficiently construct the reduced operator and right-hand side for new parameter values during the online stage. Unlike other non-intrusive ROM methods, we enforce the reduced equation during the online stage. In addition, the non-intrusive nature of our algorithm makes it straightforward and applicable to general nonlinear time-independent problems. We demonstrate its performance through several benchmark examples, including nonlinear and multiscale PDEs.

A CFD study on ultrasound-enhanced CFB riser with calcium oxide and activated coal for CO2 capture application

This numerical study explores techniques for optimizing fluidization in circulating fluidized beds (CFB), which are of significant industrial interest. Various intrusive and non-intrusive approaches have been investigated to mitigate issues such as cluster formation, core-annulus profiles, and solid back-mixing. Specifically, this research focuses on non-intrusive methods, employing ultrasonic waves to enhance solids distribution within a CFB riser by Computational Fluid Dynamics (CFD) simulations. Particulate phases, including calcium oxide and activated carbon particles, were studied alongside a gas phase of air with a 10 % carbon dioxide concentration, commonly encountered in carbon capture processes. A radial ultrasonic device featuring 20 transducers at 20 kHz and 40 kHz was affixed to the CFB to disperse particle flow. Numerical experiments within a 3-D CFD domain employed the URANS-k-ε-KTGF-EMMS mathematical model. Findings reveal that ultrasonic waves significantly influence particle motion, modifying riser flow profiles and enhancing solids distribution. While ultrasounds increase pressure drop within the transducers' region, they decrease it above, reshaping the flow profile for both particle types. Distinct outcomes are observed for calcium oxide and coal particles regarding the radial Péclet number and solids volume fraction dispersion coefficient, suggesting that ultrasounds effectively improve coal particle distribution along the riser's height.

In situ model fusion for building digital twinning

Building digital twins are crucial in holistic, synthetic, and supervisory digital environments for intelligent and energy-efficient building operations. As buildings are designed, constructed, and operated for long-term periods with uncertainties, it is essential to model, verify, extend, and calibrate the digital twin models in situ over the building life cycle, unlike digital twin modeling in the manufacturing industries. Therefore, this study proposes in situ model fusion techniques for building digital twinning, which include (1) model coupling and (2) model assembly. The first model coupling technique enables nonintrusive in situ verification and calibration for prediction models without the model observations (Y). The second model assembly enables in situ modeling or a more accurate model construction by connecting the verified models through the model coupling technique to the input layer of the target model. In the case studies conducted in a central heating system, the prediction model for the secondary-side return water temperature was modeled and calibrated without relying on the model observation data (Y), thanks to the model coupling technique. The nonintrusive model showed an accuracy with a root mean square error (RMSE) of 0.53 °C, which was comparable to the intrusive gray-box models (around 0.5 °C RMSE) calibrated using intrusive datasets (Y). Additionally, the prediction model could improve the benchmark model's performance from 0.44 °C to 0.26 °C (RMSE) using the model assembly technique. The in situ model fusion would enable and enhance nonintrusive modeling approaches, providing a reliable and extensive model environment for digital twins in the building sector.

A quadratic decoder approach to nonintrusive reduced‐order modeling of nonlinear dynamical systems

AbstractLinear projection schemes like Proper Orthogonal Decomposition can efficiently reduce the dimensions of dynamical systems but are naturally limited, e.g., for convection‐dominated problems. Nonlinear approaches have shown to outperform linear methods in terms of dimension reduction versus accuracy but, typically, come with a large computational overhead. In this work, we consider a quadratic reduction scheme which induces nonlinear structures that are well accessible to tensorized linear algebra routines. We discuss that nonintrusive approaches can be used to simultaneously reduce the complexity in the equations and propose an operator inference formulation that respects dynamics on nonlinear manifolds.

Occurrence and behavioral rhythms of the endangered Acadian redfish (Sebastes fasciatus) in the Sambro Bank (Scotian Shelf)

The genus Sebastes is a morphologically and ecologically diverse genus of rockfish characterized by high longevity, late-maturity and low natural mortality. On the northwest Atlantic continental shelf, the Acadian redfish (Sebastes fasciatus) is the most common rockfish species above 300 m depth. This species has been widely exploited resulting in the depletion or collapse of most of its stocks. Management of long-lived species with intricate life-history characteristics is challenging and requires highly integrated biological and oceanographic monitoring, which allow the identification of environmental drivers and demographic and behavioral trends. The present study uses high-temporal resolution imaging and environmental data, acquired with an autonomous lander deployed for 10-months at the Sambro Bank Sponge Conservation Area (Scotian Shelf) to elucidate S. fasciatus temporal dynamics and behavioral trends in response to near-bed environmental conditions. S. fasciatus, mostly displayed passive locomotion and static behaviors, in common with other shelf-dwelling Sebastes species. Structural complexity provided by sponges positively influenced S. fasciatus presence. Fish used sponges to avoid being dragged by bottom currents. Hydrodynamics appear to act as a synchronizing factor conditioning its swimming behavior. S. fasciatus total counts exhibited a seasonal shift in rhythm’s phase likely reflecting changes in lifestyle requirements. This study provides new insights on S. fasciatus dynamics and behavior. Nonintrusive monitoring approaches, such as the one used in this study, will be key to monitor this threatened species populations. Especially, since it is expected that S. fasciatus will experience distribution shifts to higher latitudes due to future climate stressors.

Particle mass flow determination in dust-laden supersonic flows by means of simultaneous application of optical measurement techniques

Particle mass flow rate and particle mass concentration are key parameters for describing two-phase flows, especially for particle-induced heating augmentation analysis. This work addresses the question of how accurate particle mass flow rate can be determined with three non-intrusive measurement approaches, based on shadowgraphy, particle tracking velocimetry (PTV), and scattered light intensity, in supersonic flows. In terms of shadowgraphy and PTV, the particle mass flow rate was determined by measuring individual particle characteristics, namely particle size, velocity, and density, as well as the measurement volume. The presented shadowgraphy procedure is based on the commercial LaVision DaVis software and additional shadowgraphy corrections. Multiple tests were conducted in the experimental test facility GBK of DLR with varying flow conditions, at a Mach number of 2.1, unit Reynolds number (Re∞) ranging from 5e7 1/m to 1.5e8 1/m, total temperature (T0) ranging from 303 to 544 K, and particle materials, namely Al2O3, MgO, and SiO2, in the size range of 1 to 60 µm. Particle size distributions of Al2O3 and MgO particles could be reproduced with shadowgraphy quite well, while the PTV procedure resulted in non-similar distributions. Pycnometer measurements indicated MgO particle density to be significantly lower than reference values. A DaVis parameter variation analysis resulted in a particle mass flow rate uncertainty of shadowgraphy of up to 30%. The particle mass flow rate uncertainty of PTV is approx. 76%, and the respective uncertainty of scaled PTV and scattered light intensity approach is 28%. The particle mass flow rate, measured with shadowgraphy, is 58% higher than those of the semi-axisymmetric scattered light intensity approach, which can be explained by a higher particle concentration at the injection plane.

Efficient nonlinear reduced order modeling for dynamic analysis of flat structures

In the present paper, strategies for reduced order modeling of geometrically nonlinear finite element models are investigated. Simulation-free, non-intrusive approaches are considered, which do not require access to the source code of a finite element program (e.g., proprietary knowledge). Our study focus on but is not restricted to flat structures. Reduction bases are generated using bending modes and the associated modal derivatives, which span the additional subspace needed for an adequate approximation of the geometrically nonlinear response. Moreover, the reduced nonlinear restoring forces are expressed as third order polynomials in modal coordinates. Consequently, the reduced systems can be effectively solved using time-integration schemes involving only the reduced coordinates. A bottleneck in the non-intrusive methods is typically the computational effort for precomputing the polynomial coefficients and generating the reduction basis. In this regard, we demonstrate that modal derivatives have several useful properties. In particular, the modal derivatives essentially provide all the information needed for generating the polynomial coefficients for the in-plane coordinates. For condensed systems, which ignores the inertia of the in-plane modes, we show that the modal derivatives can be used effectively for recovering the in-plane displacements. Based on these findings, we propose a methodology for generating reduced order models of geometrically nonlinear flat structures in a computationally efficient manner. Moreover, we demonstrate that the concepts extend also to curved structures. The modeling techniques are validated by means of numerical examples of solid beam models and continuously supported shell models. The computational efficiency of the proposed methodology is evaluated based on the number of static evaluations needed for identifying the polynomial coefficients, as compared to the state-of-the-art methods. Furthermore, strategies for efficient time integration are discussed and evaluated.

IoT-Based Non-Intrusive Automated Driver Drowsiness Monitoring Framework for Logistics and Public Transport Applications to Enhance Road Safety

The exponential growth in road accidents has led to a need for continuous driver monitoring to enhance road safety. Existing techniques rely on vehicle sensor-based and behavior analysis-based approaches, where the behavior analysis-based approaches are generally considered more desirable as they enable reliable detection of a more elaborate set of driver behaviors. They are categorized as intrusive and non-intrusive approaches. Unlike intrusive approaches that generally rely on constant direct human contact with sensors (physiological signals) and are sensitive to artifacts, non-intrusive approaches offer a more effective behavior monitoring using computer vision-based techniques. This paper proposes an end-to-end non-intrusive IoT-based automated framework to monitor driver behaviors, designed specifically for logistic and public transport applications. It consists of an embedded system, edge computing and cloud computing modules, and a mobile phone application, in an attempt to provide a holistic unified solution for drowsiness detection, monitoring, as well as evaluation of drivers. Drowsiness detection is based on detecting sleeping, yawning, and distraction behaviors using an image processing-based technique. To minimize the effects of latency, throughput, and packet losses, edge computing is performed using commercial off-the-shelf embedded boards. Moreover, a cloud-hosted real-time database for remote monitoring on interactive Android mobile application has been set up, where admin can add multiple drivers to get drowsiness notifications along with other useful related information for driver evaluation. An extensive experimental testing has been performed, obtaining encouraging results. An overall accuracy of 96% is achieved along with an enhanced robustness, portability, and usability of the proposed framework.

Parametrized non intrusive space-time approximation for explicit dynamic fem applications

In the following work, a benchmark of different non-intrusive model reduction approaches is performed on an explicit dynamic contact 3D-problem. The main purpose of this work is to evaluate the stability of the reduced model with respect to time along with the precision of these approaches with respect to the true solutions of interest. These solutions are the prediction of displacement and velocity fields. The precision of these approaches is also evaluated with respect to the evolution of some materials parameters. Six parameters vary in this study and we would like to predict the whole transient fast dynamic impact response with respect to each parameters. To this end, several models are trained : Proper Orthogonal Decomposition (POD) and Deep convolutional Neural Network (DcNN), in addition, a vectorized version of Interpolation in Grassman Manifolds is proposed. The benchmark performed illustrate that using DcNN’s allows to achieve the best precision and stability in predicting physical fields.

In-situ observation and calibration in building digitalization: Comparison of intrusive and nonintrusive approaches

Under digitalization-driven carbon neutrality in the building energy sector, in-situ virtual sensors could play a role in informative sensing environments and information modeling in building operations. This study proposes methodologies for the in-situ observation of virtual sensors and their in-situ calibration in building operations and digitalization. In the context of the building life cycle, in-situ observation virtual sensors are developed based on building related physics principles and design information, without physical observations, and then they are calibrated using two calibration approaches: (1) nonintrusive-indirect and (2) intrusive-direct calibrations. A comparative study is conducted to investigate the effectiveness of physics-based in-situ virtual sensors in real operation, compare the two calibration performances, and suggest recommendations for each calibration approach for real applications. According to a field study in a target district substation serving residential buildings, the in-situ observation virtual sensor for the demand-side return water temperature showed a root mean squared error (RMSE) of 0.81 °C before calibration in the heating season (112 days). The RMSEs of 0.61 and 0.55 °C were found for the nonintrusive-indirect and intrusive-direct calibrations in the representative case, respectively. These results showed the effectiveness of nonintrusive-indirect calibration, even though the target observations were unknown in the calibration. This study recommends obtaining informative datasets for improved intrusive calibration. In the case study, calibration with the informative dataset for 1 day (RMSE of 0.59 °C) was superior to that of the average RMSE for nonintrusive calibration. The current energy patterns could be a basis for deciding whether the next few days are suitable for obtaining informative intrusive datasets, resulting in better calibration accuracy or shortening of the required datasets.

Nonintrusive Perceptual Audio Quality Assessment for User-Generated Content Using Deep Learning

With the boom of social media communication, teleconferencing, and online classes, audiovisual communication over bandwidth strained networks has become an integral part of our lives. Consequently, the growing demand for the quality of experience necessitates developing algorithms to measure and enrich user experience. Prior studies have mainly focused on assessing speech quality and intelligibility with reference to audio quality assessment, while other categories in user-generated multimedia (UGM) are less explored. Moreover, frequency-domain properties of speech and UGM audio are significantly different from each other. Furthermore, there is a lack of a standard dataset for the quality assessment of UGM. Considering these limitations, in this article, we first develop the IIT-JMU-UGM audio dataset consisting of 1150 audio clips, with diverse context, content, and types of degradation commonly observed in real-world scenarios and annotated with the subjective quality scores. Finally, we propose a non-intrusive audio quality assessment metric using a stacked gated-recurrent-unit-based deep learning framework. The proposed model outperforms several baseline methods, including state-of-the-art non-intrusive and intrusive approaches. The resulting Pearson’s correlation coefficient of 0.834 indicates that the proposed method efficiently mirrors human auditory perception.

Non-intrusive estimation of model error and discrepancy in dynamics models

Errors in formulating the system physics model, due to inability to rigorously account for nonlinear behavior, boundary conditions, and multi-component and multi-physics interactions result in discrepancies between model predictions of system responses and the corresponding experimentally measured values. In our previous work, we have developed a Bayesian state estimation-based framework for the estimation of these model errors, and the subsequent prediction of model discrepancies at unmeasured locations, for untested inputs, and for untested configurations. In this approach, we represent model errors as external inputs in the dynamic system model, and estimate them using Bayesian state estimation. However, this approach is intrusive in the sense that it requires access to the governing system of equations, the ability to modify the system model by introducing an external system input, and the ability to represent system inputs and responses as probabilistic quantities. In this paper, we develop semi-intrusive and non-intrusive approaches to extend the model error estimation to black-box models. The semi-intrusive method does not need access to the governing equations, but requires the ability to interrupt the simulation at any time instant, update the system responses at that moment, and resume the simulation using the updated value of system responses. For situations where this is not permitted, we propose a fully non-intrusive approach where, a surrogate model is constructed and used in the Bayesian state estimation procedure. The proposed approaches are illustrated with three examples: structural deformation of a Timoshenko beam, heat transfer across a rigid bar, and deformation of a Timoshenko beam subjected to heating.

Numerical Assessment of a Nonintrusive Surrogate Model Based on Recurrent Neural Networks and Proper Orthogonal Decomposition: Rayleigh–Bénard Convection

Recent developments in diagnostic and computing technologies offer to leverage numerous forms of nonintrusive modelling approaches from data where machine learning can be used to build computationally cheap and accurate surrogate models. To this end, we present a nonlinear proper orthogonal decomposition (POD) framework, denoted as NLPOD, to forge a nonintrusive reduced-order model for the Boussinesq equations. In our NLPOD approach, we first employ the POD procedure to obtain a set of global modes to build a linear-fit latent space and utilise an autoencoder network to compress the projection of this latent space through a nonlinear unsupervised mapping of POD coefficients. Then, long short-term memory (LSTM) neural network architecture is utilised to discover temporal patterns in this low-rank manifold. While performing a detailed sensitivity analysis for hyperparameters of the LSTM model, the trade-off between accuracy and efficiency is systematically analysed for solving a canonical Rayleigh–Bénard convection system.

A Survey on Anti-Spoofing Methods for Facial Recognition

Despite significant development in facial recognition (FR), current FR systems are exposed to spoofing attacks like printed photo attacks, 3D mask attacks, video replay attacks, and many more. Several anti-spoofing approaches have been proposed to assess whether the person in front of the camera is real or fake. Developing effective protection mechanisms against these threats is a challenging task. This paper gives a brief overview of various presentation attack detection (PAD) techniques, which are categorized into intrusive and non-intrusive approaches. Each technique is examined in terms of its execution, benefits, and drawbacks and also provides information on modern anti-spoofing techniques.

Machine learning based optimization for interval uncertainty propagation

Two non-intrusive uncertainty propagation approaches are proposed for the performance analysis of engineering systems described by expensive-to-evaluate deterministic computer models with parameters defined as interval variables. These approaches employ a machine learning based optimization strategy, the so-called Bayesian optimization, for evaluating the upper and lower bounds of a generic response variable over the set of possible responses obtained when each interval variable varies independently over its range. The lack of knowledge caused by not evaluating the response function for all the possible combinations of the interval variables is accounted for by developing a probabilistic description of the response variable itself by using a Gaussian Process regression model. An iterative procedure is developed for selecting a small number of simulations to be evaluated for updating this statistical model by using well-established acquisition functions and to assess the response bounds. In both approaches, an initial training dataset is defined. While one approach builds iteratively two distinct training datasets for evaluating separately the upper and lower bounds of the response variable, the other one builds iteratively a single training dataset. Consequently, the two approaches will produce different bound estimates at each iteration. The upper and lower response bounds are expressed as point estimates obtained from the mean function of the posterior distribution. Moreover, a confidence interval on each estimate is provided for effectively communicating to engineers when these estimates are obtained at a combination of the interval variables for which no deterministic simulation has been run. Finally, two metrics are proposed to define conditions for assessing if the predicted bound estimates can be considered satisfactory. The applicability of these two approaches is illustrated with two numerical applications, one focusing on vibration and the other on vibro-acoustics.

DriverRep: Driver identification through driving behavior embeddings

Driver identification has emerged as an active field of study to further personalize the integrated advanced driver-assistance systems into intelligent vehicles, provide security and safety for ride-hailing services, and prevent auto theft. Several studies, within recent years, have investigated non-intrusive identification approaches, focusing on driving behavior analysis to characterize drivers' driving behavior, in which a considerable volume of labeled data is required. This paper develops a deep learning architecture, namely DriverRep, to extract the latent representations associated with each individual, called driver embeddings. These embeddings represent the unique driving characteristics of drivers. To this aim, we introduce a fully unsupervised triplet loss that selects triplet samples from data in an unsupervised manner and extracts the embeddings using our proposed stacked encoder architecture. Dilated causal convolutions are used to make residual blocks of the encoder. To perform the task of driver identification, we leverage the classification accuracy of SVM on top of the obtained driver embeddings. The evaluation results over two datasets, each of which contains ten drivers, reveal that the DriverRep can successfully capture the underlying features within the data and outperform benchmark driver identification schemes, obtaining an average accuracy of 94.7% and 96% for two-way and three-way identification, respectively. We also investigate the ability of the DriverRep in handling sparsely labeled data. The results represent substantial improvements in comparison with the supervised approaches when applied to highly sparsely labeled data.

Uncertainty Quantification of Differential Algebraic Equations Using Polynomial Chaos

Abstract The focus of this paper is on the use of polynomial chaos (PC) for developing surrogate models for differential algebraic equations (DAEs) with time-invariant uncertainties. Intrusive and nonintrusive approaches to synthesize PC surrogate models are presented including the use of Lagrange interpolation polynomials as basis functions. Unlike ordinary differential equations (ODEs), if the algebraic constraints are a function of the stochastic variable, some initial conditions of the DAEs are also random. A benchmark RLC circuit which is used as a benchmark for linear models is used to illustrate the development of a PC-based surrogate model. A nonlinear example of a simple pendulum also serves as a benchmark to illustrate the potential of the proposed approach. Statistics of the results of the PC models are validated using Monte Carlo (MC) simulations in addition to estimating the evolving probably density functions (PDFs) of the states of the pendulum.