Michigan Technological University
  • Houghton, Michigan, United States
Recent publications
We present a data-driven method based on deep learning for identifying nonlinear normal modes of unknown nonlinear dynamical systems using response data only. We leverage the modeling capacity of deep neural networks to identify the forward and inverse nonlinear modal transformations and the associated modal dynamics evolution. We test the method on Duffing systems with cubic nonlinearity and observe that the identified NNMs with invariant manifolds from response data agree with those analytical or numerical ones using closed-form equations.
In this study, we investigate the performance of data-driven Koopman operator and nonlinear normal mode (NNM) on predictive modeling of nonlinear dynamical systems using a physics-constrained deep learning approach. Two physics-constrained deep autoencoders are proposed: one to identify eigenfunction of Koopman operator and the other to identify nonlinear modal transformation function of NNMs, respectively, from the response data only. Koopman operator aims to linearize nonlinear dynamics at the cost of infinite dimensions, while NNM aims to capture invariance properties of dynamics with the same dimension as original system. We conduct numerical study on nonlinear systems with various levels of nonlinearity and observe that NNM representation has higher accuracy than Koopman autoencoder with same dimension of feature coordinates.
This paper analyzes the influence of anti-aliasing filters (AAFs) on traveling wave (TW)-based fault location (TWFL) methods. To do so, the Alternative Transients Program (ATP) is used to simulate several fault scenarios on a 230 kV/60 Hz test power system, allowing the assessment of AAFs influence when different TW filters and TWFL algorithms are used. The main goal is to clarify how much “villain” are the AAFs for TWFL methods if their cutoff frequencies are varied in noisy environments. By doing so, the most relevant spectrum band for TWFL solutions is investigated, addressing whether the application of AAFs with reduced cutoff frequencies would be critical or not. Although most people may think that the answer to this question is obvious, the obtained results show evidences that some solutions monitor transients within spectrum ranges wider than those they indeed require. Moreover, although AAFs can relevantly influence the performance of TWFL methods, it is proven that the application of AAFs with reduced cutoff frequencies is not necessarily critical, but rather, it depends on the used TW filters and TWFL methods, such that it can be beneficial when electrical noise is present in monitored signals.
This study numerically investigated the thermal Marangoni flow and heat transfer characteristics of an evaporating droplet. Uniform and local heating methods can control the internal flow patterns of a droplet during evaporation. The present study applied the dynamic mesh method to simulate the behaviors of the liquid-air interface during evaporation. The results revealed that the flow transition inside the droplet appeared in the early stages of evaporation and occurred owing to the temperature variation at the liquid-air interface; these variations eventually yielded surface tension gradients. Moreover, nonuniform evaporation fluxes caused capillary flows that moved from the center of the droplet to the contact line along the substrate. The surface tension gradient along the liquid-air interface had a dominant effect on the internal flow, which induced the thermal Marangoni flow. For the local heating cases, different flow patterns appeared, as compared with those that appeared from uniform heating. The flow directions changed according to the local heating conditions owing to the difference in the local surface tensions at the liquid-air interface.
Providing sufficient benefits to local people can be an important component of effective and equitable conservation, especially where local communities face substantial opportunity costs or disbenefits from conservation. However, the distribution of benefits to local people is often inadequate or inequitable. In this study we investigated the heterogeneity in the extent to which people living near Hwange National Park (HNP), Zimbabwe, perceive benefit from the presence of the park. Specifically, we examined the relationships between a diverse set of candidate predictor variables and perceived benefit from HNP. Our candidate predictor variables broadly relate to personal assets, social capital, value orientation, fear of lions, and belief and participation in human-wildlife conflict mitigation schemes. One third of respondents reported that their household experienced at least some benefits from HNP. Of all respondents, 6% perceived their household to benefit strongly from HNP and 2% very strongly. Livestock loss to wildlife was the most important factor for predicting perceived benefit, with those suffering more loss less likely to perceive benefit. Multiple demographic factors predicted perceived benefit with, for instance, older people and those with less education perceiving less benefit. Employment in conservation-related work positively affected perceived benefit, whereas fear of lions had a negative impact. Social capital appeared to have a positive influence on perceived benefit from HNP. The relationship between social capital and perceived benefit was positive and plateauing, which suggests that social capital is especially impactful on the benefit perceived by individuals reporting the least social capital. We also found a positive association between belief in compensation schemes and perceived benefit from HNP. We posit hypotheses for this association but are unable to determine the underlying drivers of this relationship. Finally, participation in the community guardians programme, a human-lion conflict mitigation programme, was positively related to perceived benefit from HNP. Thus, our findings emphasise the value of considering a diverse array of factors when investigating park-people relationships and yield insights for improving the equitability of conservation in and around HNP and similar systems.
Alumina oxide supported nickel (Ni/Al2O3) catalysts generally suffer from fast deactivation caused by coking and formation of nickel aluminate (NiAl2O4) in dry reforming of methane (DRM). Herein, for the first time, a self-stabilization mechanism of the Ni/Al2O3 catalyst (with 0.1 wt% Ni loading) was revealed and effectively applied for DRM. Namely, the conversion of catalytically active Ni species into catalytically inert NiAl2O4 spinel in DRM over the Ni/γ-Al2O3 catalyst could be mitigated by repeated reduction-reaction treatments owing to the increasing amount of Ni located on the NiAl2O4 isolation layer rather than the reactive γ-Al2O3. The self-stabilization could be achieved over Ni/α-Al2O3 as well, even with a faster rate, since the NiAl2O4 isolation layer can be directly formed in the first reduction-reaction cycle due to its small surface area and weak metal-support interaction. These observations not only highlight the importance of an isolation layer for protecting the catalyst from deactivation, but also provide a novel and efficient self-stabilization approach for catalytic DRM. Graphical Abstract
Land use and land cover changes (LULCCs) in mountainous areas may increase the susceptibility to landslides due to modifications of topography, vegetation, and material characteristics. Understanding the relation between LULCCs and landslide occurrences is important for landslide prevention and land resources management. In this study, these changes were analyzed for the landslides that were triggered during the 2018 monsoon event in Kerala, India. The changes in land use and land cover (LULC) that took place in a period of eight years prior to the 2018 event were analyzed for 4,728 landslide initiation points in the entire state, and for a subset of 2,223 landslides in the most affected district of Idukki. Apart from this, the initiation points were compared to those in steep areas and landslide susceptible zones. For these comparisons, we used LULC datasets for the period between 2000 and 2018 that were obtained from national organizations, derived from satellite image classification and visual interpretation. The LULC datasets lacked coherent classification schemes, so a standardization of LULC types was made. The results of these comparisons reveal that more than half of the landslides (58%) occurred in densely vegetated areas, and that 50% of the landslides that caused damage to buildings and roads originated in forest plantations, followed by built-up areas (25%). For most of the landslide locations (90% in Kerala and 83% in Idukki) there was no noticeable change in major LULC in the period prior to the event. Results of this study indicate that LULCCs for the the period 2010-2018 had less influence on the landslides occurred in 2018.
This paper develops an end-to-end ECG signal classification algorithm based on a novel segmentation strategy and 1D Convolutional Neural Networks (CNN) to aid the classification of ECG signals and alleviate the workload of physicians. The ECG segmentation strategy named R-R-R strategy (i.e., retaining ECG data between the R peaks just before and after the current R peak) is used for segmenting the original ECG data into segments to train and test the 1D CNN models. The novel strategy mimics physicians in scanning ECG to a greater extent, and maximizes the inherent information of ECG segments for diagnosis. The performance of the proposed end to end ECG signal classification algorithm was verified with the ECG signals from 48 records in the MIT-BIH arrhythmia database. When the heartbeat types were divided into the five classes recommended by clinicians, i.e., normal beat, left bundle branch block beat, right bundle branch block beat, premature ventricular contraction, and paced beat, the classification accuracy, the area under the curve (AUC), the sensitivity, and the F1-score achieved by the proposed model were 0.9924, 0.9994, 0.99 and 0.99, respectively. When the heartbeat types were divided into six classes recommended by clinicians, i.e., normal beat, left bundle branch block beat, right bundle branch block beat, premature ventricular contraction, paced beat and other beats, the beat classification accuracy, the AUC, the sensitivity, and the F1-score achieved by the model reached 0.9702, 0.9966, 0.97, and 0.97, respectively. When the heartbeat types were divided into five classes recommended by the Association for Advancement of Medical Instrumentation (AAMI), i.e., normal beat, supraventricular ectopic beat, ventricular ectopic beat, fusion beat, and unknown beat, the beat classification accuracy, the sensitivity, and the F1-score were 0.9745, 0.97, and 0.97, respectively. Experimental results show that the proposed method achieves better performance than the state-of-the-art methods.
Tumor Necrosis Factor (TNF)-α is a proinflammatory cytokine (PIC) and has been implicated in a variety of illness including cardiovascular disease. The current study investigated the inflammatory response trigged by TNFα in both cultured brain neurons and the hypothalamic paraventricular nucleus (PVN), a key cardiovascular relevant brain area, of the Sprague Dawley (SD) rats. Our results demonstrated that TNFα treatment induces a dose- and time-dependent increase in mRNA expression of PICs including Interleukin (IL)-1β and Interleukin-6 (IL6); chemokines including C–C Motif Chemokine Ligand 5 (CCL5) and C–C Motif Chemokine Ligand 12 (CCL12), inducible nitric oxide synthase (iNOS), as well as transcription factor NF-kB in cultured brain neurons from neonatal SD rats. Consistent with this finding, immunostaining shows that TNFα treatment increases immunoreactivity of IL1β, CCL5, iNOS and stimulates activation or expression of NF-kB, in both cultured brain neurons and the PVN of adult SD rats. We further compared mRNA expression of the aforementioned genes in basal level as well as in response to TNFα challenge between SD rats and Dahl Salt-sensitive (Dahl-S) rats, an animal model of salt-sensitive hypertension. Dahl-S brain neurons presented higher baseline levels as well as greater response to TNFα challenge in mRNA expression of CCL5, iNOS and IL1β. Furthermore, central administration of TNFα caused significant higher response in CCL12 in the PVN of Dahl-S rats. The increased inflammatory response to TNFα in Dahl-S rats may be indicative of an underlying mechanism for enhanced pressor reactivity to salt intake in the Dahl-S rat model.
Stay-cables in the cable-stayed bridge are the most vital components as they carry the bridge deck’s load and transmit the force to the bridge pylons. However, dynamics loads due to vortex-induced vibration, ambient wind excitation, and even vehicular vibration cause fatigue in the stay-cable. Hence continuous real-time performance monitoring of such cables is necessary for maintenance to avoid any kind of damage to the cable. Wireless sensors are contact-based sensor that provide accurate measurement, and it does not involve any wiring cost like conventional wired sensors. Monitoring cable health using such wireless sensors is a good choice provided packet loss (which occurs while transmitting the measured data to the base station) that invariably occurs is addressed by data processing. Such discontinuity in data (due to packet loss) may interrupt the real-time/online cable health monitoring process - depending on the window length of the data loss. In general, online health monitoring using multiple sensors reduces the estimation errors. In this paper, we propose a framework that takes the wireless sensor data as the input, then reconstructs the packet lost samples (if any), and finally, provides a real-time tension estimation as an output. The novel framework, first adopts compressive sensing algorithm to reconstruct the data due to packet loss. Subsequently, we synthesize the reconstructed responses from multiple sensors to estimate the real-time frequency variation using Blind Source Separation (BSS) Technique. As the cable response due to ambient vibration contains a large number of modes, the dominant modal response or the corresponding dominant frequency is estimated from very few measurements using a variant of the BSS technique named Sparse Component Analysis (SCA). Finally, real-time cable tension is estimated from the frequency variation using the taut-string theory. The proposed technique is applied to a real full-scale cable-stayed bridge. The mean tension obtained from the framework is comparable with the cable’s actual design tension. The accurate estimation of real-time stay-cable tension by the proposed algorithm shows great potential in the field of structural health monitoring.
In this paper, we apply local discontinuous Galerkin (LDG) methods to compressible wormhole propagation with Darcy–Forchheimer model. We consider two time integrations up to second-order accuracy and prove the stability of the fully-discrete schemes. There are several difficulties. Firstly, different from most previous works discussing stability of wormhole propagations, we use LDG methods and have to deal with the inter-element discontinuities, leading to more complicated theoretical analysis. Secondly, in most previous stability analysis of LDG methods, a key step is to construct the relationship between the derivatives of the primitive variable and the auxiliary variables. This idea works for linear problems. However, our system is highly nonlinear and all the variables are coupled together. As an alternative, we will introduce a new auxiliary variable containing both the convection and diffusion terms. Thirdly, we have to control the change of the porosity during time evolution to obtain physically relevant numerical approximations and uniform upper bounds. Fourthly, to handle the time level mismatch of the spatial discretization due to the time integrations, we will construct a special second-order time method. Finally, to handle the complexity due to the Forchheimer term, we extrapolate some non-essential variables to linearize the coupled system, avoiding complicated iterations. To the best knowledge of the authors, this is the first scheme with time accuracy greater than one discussing stability for wormhole propagations. Moreover, we will prove the optimal error estimates of the schemes under mild time step restrictions. Numerical experiments are also given to verify the theoretical results.
Wet deposition has been well recognized to be affected by species concentration and precipitation; nevertheless, the regimes in the controlling factor of concentration or precipitation have not yet been clarified. Using a trace element, selenium (Se), with dual effects on human health as a testbed, we first reproduce the spatial distribution of atmospheric Se concentrations and wet deposition fluxes through GEOS-Chem on a global scale, and examine the spatial patterns and relative importance of anthropogenic emissions vs. natural emissions over various regions around the world. We find that over most Northern Hemisphere continental regions, anthropogenic emissions are the dominant source for atmospheric Se concentration and deposition, while it is dominated by natural sources in the other areas. Nested grid simulations covering China and the continental United States are further conducted. The factors (i.e., Se concentration and precipitation) controlling the wet deposition flux of atmospheric Se are analyzed in detail, through the construction of wet deposition-concentration-precipitation (WETD-C-P) diagram for two regions (mainland China and the continental United States) based on the monthly results. The two regions show distinctive features, reflecting the different spatial patterns of Se emissions and precipitation. Both Se emissions and precipitation are higher in the eastern United States than that in the western United States. In contrast, the emissions and precipitation in northern and southern China show dipole features with stronger emissions over the northern side and higher precipitation on the southern side. We further investigate the impacts of future emission changes in China on atmospheric Se deposition and its sensitivity to emissions and precipitation, revealing a modulation of regime shifts, i.e., from the precipitation dominant regime to the concurrent governance of both precipitation and emissions. The proposed WETD-C-P relationship is useful in elucidating the regime and factors governing the spatial and temporal variations in wet deposition.
Boron-based 2D monolayers have attracted tremendous interest due to their unique physical and chemical properties. In this paper, we report novel pentagonal monolayers, B2S and B2Se, which are predicted to be energetically, dynamically, and thermally stable based on density functional theory. At the HSE06 level of theory, they exhibit a moderate indirect bandgap of (e.g., 1.82 eV for Penta-B2S and 1.94 eV for Penta-B2Se). Strain-induced indirect-to-direct bandgap transition, high hole mobility (~103 Cm2V-1S-1) and strong optical absorption (α ~105 Cm-1) in the visible region are observed for these monolayers. Moreover, the electronic band structures and optical spectra are tunable by mechanical strains suggesting their visible light-harvesting capabilities for optoelectronic applications. In this way, the pentagonal family of 2D materials is now expanded to include boron-containing photocatalytic materials for water splitting applications.
The effect of freeze–thaw cycles on the pavement performance of SBS modified asphalt mixture (SBSMA) and SBS/crumb rubber composite modified asphalt mixture (CCRMA) was studied using a triaxial repeated creep test, semicircular bending fatigue test based on digital image correlation (DIC), and thermal stress restrained specimen test. The mechanical model of the residual strain (RS) of the asphalt mixtures is established, and the residual viscous flow strain (εRvf,N) and residual viscoelastic strain (εRve,N) are separated. Based on the horizontal strain (Exx) of DIC, a new mesofatigue damage variable (D) is proposed. According to the characteristic curve of D, the antifatigue damage parameter (AFDP) is established and the correlation between AFDP and fatigue life (Nf) is analyzed. Simultaneously, a calculation model of the relaxation temperature stress (σRts(T)) of the asphalt mixture at different cooling rates (v) is established based on the finite element method. Results showed that the growth rates of RS, εRvf,N, and D of both asphalt mixtures accelerate, AFDP decreases, and freeze–break temperature (TD) rises with increasing freeze–thaw cycles. However, the growth rates of RS, εRvf,N, and D of CCRMA are always slower than SBSMA, the AFDP is larger than SBSMA, TD is lower than SBSMA, εRve,N is a fixed value under specific conditions, and a good correlation exists between the AFDP and Nf. The pavement performance of the asphalt mixtures worsens after the freeze–thaw cycles, and the influence of a salt freeze–thaw cycles is greater than that of a water freeze–thaw cycles. Moreover, CCRMA has better high-temperature deformation resistance, fatigue damage resistance, and low-temperature cracking resistance than SBSMA. The conversion point temperature (TZ) increases with the acceleration of v, and the σRts(T) calculation model can accurately reflect the σRts(T) of the asphalt mixture at different v.
In winter, icing greatly reduces the friction performance of pavement and brings a high safety hazard to road traffic. We prepared a superhydrophobic coating (SHC) with a microwave heating function to minimize pavement icing risk. The static contact angle of the water droplet on the SHC is 161.2°. The freezing time of water droplets on asphalt mixture coated with SHC is 1.63 times longer than that of asphalt mixture without coating. SHC can significantly delay the freezing process of water. The normal adhesion strength of 10 mm ice layer on the asphalt mixture coated with the SHC at −5 °C is only 14.3 % of that on the asphalt mixture without coating. It confirms that the SHC dramatically reduces the normal adhesion strength of the ice layer on asphalt pavement. In addition, the microwave heating rate of the asphalt mixture coated with SHC is 48.37 % faster than that of the asphalt mixture without coating. Carbon nanotube particles directly contact the ice on the SHC surface. Besides, carbon nanotube particles also have high microwave heat conversion efficiency and high thermal conductivity. The ice layer on the surface of the asphalt mixture coated with SHC can be melted and removed within 80 sec, which is 42.86 % faster than the ice removal time of asphalt mixture without coating. Carbon nanotube particles on the SHC surface were directly heated by microwaves and melted the ice layer rapidly. The acrylic coating on the asphalt mixture surface has an excellent thermal insulation performance. The load wheel rolling experiment was conducted to simulate the traffic wheel friction. The static contact angle of the water droplet on the SHC keeps at 147°, and the microwave heating performance of the SHC has no significant changes before and after 500 rolling times. This implies that the SHC has sustainable superhydrophobic properties and can melt the ice layer under microwave heating even after the traffic wheel friction. The SHC is of great significance for improving the driving safety of asphalt pavement.
Pavement compaction cannot be neglected during the motorway manufacture stage because it can determine pavement service quality and durability. Concerning the compaction scenario, the paving compaction is responsible for offering the preliminary strength of the pavement. Ignoring paving compaction quality control can lead to over compaction. This paper introduces an integral system to study and simulate the paving compaction of asphalt motorways in Discrete Element Model two-dimensional (DEM2D). This method includes the whole procedure from aggregate image acquisition database establishment to the DEM2D simulation of paving compaction. To this end, this study fulfils the creation of the aggregate database applied in DEM via the Aggregate Image Measuring System (AIMS) method. In addition, the artificial intelligent (AI) technology called Generative Adversarial Networks (GANs) method is proposed to expand the developed DEM aggregate database. Three different approaches are applied to calibrate the accuracy of the extended database. According to the aggregate database, the pavement paving compaction with different aggregate gradations can be simulated in DEM2D.
Forum papers are thought-provoking opinion pieces or essays founded in fact, sometimes containing speculation, on a civil engineering topic of general interest and relevance to the readership of the journal. The views expressed in this Forum article do not necessarily reflect the views of ASCE or the Editorial Board of the journal.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
3,417 members
Dana Johnson
  • College of Business
Nancy HF French
  • Michigan Tech Research Institute
Bowen Li
  • Department of Materials Science & Engineering
William H Cooke
  • Department of Kinesiology and Integrative Physiology
Sean J Kirkpatrick
  • Department of Biomedical Engineering
1400 Townsend Dr, 49931, Houghton, Michigan, United States