Efficient Modeling Approach Research Articles

Simulated tissue growth in tetragonal lattices with mechanical stiffness tuned for bone tissue engineering

Bone tissue engineering approaches have recently begun considering 3D printed lattices as viable scaffold solutions due to their highly tunable geometries and mechanical efficiency. However, scaffold design remains challenging due to the numerous biological and mechanical trade-offs related to lattice geometry. Here, we investigate novel tetragonal unit cell designs by independently adjusting unit cell height and width to find scaffolds with improved tissue growth while maintaining suitable scaffold mechanical properties for bone tissue engineering. Lattice tissue growth behavior is evaluated using a curvature-based growth model while elastic modulus is evaluated with finite element analysis. Computationally efficient modeling approaches are implemented to facilitate bulk analysis of lattice design trade-offs using design maps for biological and mechanical functionalities in relation to unit cell height and width for two contrasting unit cell topologies. Newly designed tetragonal lattices demonstrate higher tissue growth per unit volume and advantageous stiffness in preferred directions compared to cubically symmetric unit cells. When lattice beam diameter is fixed to 200 μm, Tetra and BC-Tetra lattices with elastic moduli of 200 MPa–400 MPa are compared for squashed, cubic, and stretched topologies. Squashed Tetra lattices demonstrated higher growth rates and growth densities compared to symmetrically cubic lattices. BC-Tetra lattices with the same range of elastic moduli show squashed lattices tend to achieve higher growth rates, whereas stretched lattices promote higher growth density. The results suggest tetragonal unit cells provide favorable properties for biological and mechanical tailoring, therefore enabling new strategies for diverse patient needs and applications in regenerative medicine.

Application of an Accurate and Efficient Modeling Approach to a Multiscale Fractured Reservoir in South China Sea

Carbonate reservoirs in the South China Sea mostly contain natural fractures with various length scales and different intensities, which causes great challenges in efficient reservoir modeling and flow simulation. Existing efforts based on dual-porosity and dual-permeability models could not reflect the characteristics of production data in certain wells. To accurately and efficiently characterize multiscale fractures, a hybrid fracture characterization method is proposed. Firstly, fractures are divided into two types according to the geometrical size and interpretation approach. Then, small-scale fractures, characterized mainly by image log interpretations, are modeled by the traditional dual-porosity/dual-permeability (DP) method. And large-scale fractures, which are characterized by seismic interpretations and dominate the flow regime, are modeled by the embedded discrete fracture method (EDFM) to achieve both accuracy and efficiency. Lastly, transmissibilities among these three types of grid mediums are calculated to generate the hybrid DP+EDFM model for flow simulation. The proposed approach is applied to a carbonate, fractured reservoir in the South China Sea. The overall procedure is fast and reliable, and water cut matches of both field and specific wells are dramatically improved. Comparing the simulation results with the conventional DP model, the proposed approach yields much more accurate predictions on rapid water breakthrough and high water cut in fractured reservoirs.

Unraveling a natural protease inhibitor from marine Streptomyces griseoincarnatus HK12 active against Chikungunya virus

Proteases play an indispensable role in the life cycles of several life-threatening organisms such as the ones causing malaria, cancer and AIDS. A targeted blockade of these enzymes could be an efficient approach for drug modeling against these causative agents. Our study was directed towards the extraction and characterization of a protease inhibitor having activity against Chikungunya virus (CHIKV). A protein-based protease inhibitor (PI) in Streptomyces griseoincarnatus HK12 with anti-viral activity against CHIKV was revealed when screened against two major proteases, papain and trypsin. The PI was efficiently extracted at 60 % ammonium sulfate saturation and purified by ion-exchange chromatography (CM-Sepharose) at 300 mM NaCl elution followed by SDS-PAGE (10 %). The protein was characterized by denaturing SDS-PAGE, reverse zymography, and MALDI-TOF peptide mass fingerprinting. The protein-based PI was studied to have a high molecular weight of 66−70 kDA. The PI was tested to supress the supress cytopathic effects (CPE) exerted by the clinically isolated virus in BHK21 cells. This was used as a measure to determine the antiviral activity. The PI exerted significant effects with an effective concentration calculated as EC50 11.21 μg/mL. The protein was found to be reported as the first of its kind which also stands out to be the first a natural protease inhibitor against the treatment of the chikungunya virus.

SVD-Krylov based Sparsity-preserving Techniques for Riccati-based Feedback Stabilization of Unstable Power System Models

We propose an efficient sparsity-preserving reduced-order modelling approach for index-1 descriptor systems extracted from large-scale power system models through two-sided projection techniques. The projectors are configured by utilizing Gramian based singular value decomposition (SVD) and Krylov subspace-based reduced-order modelling. The left projector is attained from the observability Gramian of the system by the low-rank alternating direction implicit (LR-ADI) technique and the right projector is attained by the iterative rational Krylov algorithm (IRKA). The classical LR-ADI technique is not suitable for solving Riccati equations and it demands high computation time for convergence. Besides, in most of the cases, reduced-order models achieved by the basic IRKA are not stable and the Riccati equations connected to them have no finite solution. Moreover, the conventional LR-ADI and IRKA approach not preserves the sparse form of the index-1 descriptor systems, which is an essential requirement for feasible simulations. To overcome those drawbacks, the fitting of LR-ADI and IRKA based projectors from left and right sides, respectively, desired reduced-order systems attained. So that, finite solution of low-rank Riccati equations, and corresponding feedback matrix can be executed. Using the mechanism of inverse projection, the Riccati-based optimal feedback matrix can be computed to stabilize the unstable power system models. The proposed approach will maintain minimized ℌ2 -norm of the error system for reduced-order models of the target models.

Fatigue reliability assessment of steel bridges considering spatial correlation in system evaluation

Fatigue is among the most critical forms of deterioration damage that occurs to steel bridges. The performance of steel bridges, which may be seriously affected due to fatigue, should be assessed and predicted. However, the code-based reliability assessment only provides a clear criterion for single details. The whole bridge or a splitting system from the bridge may contain many details which brings system impact on the fatigue assessment. In this paper the system reliability of steel bridges is addressed with focus on the influence of spatial correlations between similar fatigue sensitive details. The study is conducted on a typical riveted railway bridge, from which a subsystem is selected and modelled as a weakest link system. It is assumed that all details in the system are equicorrelated. An efficient modeling approach has been developed to quantify the correlation and help estimate the system reliability analytically. A parametric study was performed to determine the system effects on the fatigue reliability assessment and identify the decisive variables.

Data-driven interval credibility constrained quadratic programming model for water quality management under uncertainty

Although integrated simulation-optimization modeling can provide a comprehensive and reliable analysis for water quality management (WQM), it is usually not easy to implement in practice. This study proposed a new efficient simulation-optimization modeling approach by leveraging the power of data-driven modeling, to support WQM under various uncertainties. A water quality simulation model is integrated with the optimization model, and then substituted by a series of numerical surrogate models based on inexact linear regression. The transformation can significantly reduce the computational burden and make it possible to implement uncertainty quantification through hybrid inexact programming. The proposed model incorporates interval quadratic programming and credibility constrained programming to deal with nonlinearity and various uncertainties associated with the management system. The proposed approach is applied to a real case study of the Grand River watershed in Canada for controlling phosphorus concentration in river water. The Grand River Simulation Model (GRSM) is employed as the physical simulation model to estimate the total phosphorus concentration in the river. Interval solutions under different confidence levels of violating the effluent standards were obtained, which can be used to generate optimal phosphorus control strategies. The results indicate the proposed data-driven interval credibility constrained quadratic programming (DICCQP) model is able to provide reliable and robust solutions for WQM by considering nonlinearity and various uncertainties while maintaining a high computational efficiency. The proposed new framework can be extended and applied to the other watersheds. The high efficiency of the proposed model makes it possible to solve large-scale complex water quality management and planning problems.

Seismic Lithofacies Distribution Modeling Using the Single Normal Equation Simulation (SNESIM) Algorithm of Multiple-Point Geostatistics in Upper Assam Basin, India

Recently, multiple-point geostatistical simulation gained much attention for its role in spatial reservoir characterization/modeling in geosciences. Accurate lithofacies modeling is a critical step in the characterization of complex geological reservoirs. In this study, multiple-point facies geostatistics based on the SNESIM algorithm integrated with the seismic modeling technique is used as an efficient reservoir modeling approach for lithofacies modeling of the fluvial Tipam formation in the Upper Assam Basin, India. The Tipam formation acts as a potential reservoir rock in the Upper Assam Basin, India. Due to the basin geological complexity and limitation in seismic resolution, many discontinuities in depositional channels in this fluvial depositional environment have been identified using conventional lithofacies mapping. This study combines three techniques to reproduce continuity of the lithofacies for better reservoir modeling. The first is simultaneous prestack inversion for inverting prestack gathers with angle-dependent wavelets into seismic attributes. A cross-plot of P-impedance and VP/VS ratio from well-log data was used to classify the different reservoir lithofacies such as hydrocarbon sand, brine sand, and shale. The second is the Bayesian approach that incorporates probability density functions (PDFs) of non -parametric statistical classification with seismic attributes for converting the seismic attributes into lithofacies volume and the probability volumes of each type lithofacies. The third technique is multiple-point geostatistical simulation (MPS) using the Single Normal equation Simulation (SNESIM) algorithm applied to training images and probability volumes as constraints for a better lithofacies model. These integrated study results proved that MPS could improve reservoir lithofacies characterization.

Computationally Efficient Forward Operator for Photoacoustic Tomography Based on Coordinate Transformations.

Photoacoustic tomography (PAT) is an imaging modality that utilizes the photoacoustic effect. In PAT, a photoacoustic image is computed from measured data by modeling ultrasound propagation in the imaged domain and solving an inverse problem utilizing a discrete forward operator. However, in realistic measurement geometries with several ultrasound transducers and relatively large imaging volume, an explicit formation and use of the forward operator can be computationally prohibitively expensive. In this work, we propose a transformation-based approach for efficient modeling of photoacoustic signals and reconstruction of photoacoustic images. In the approach, the forward operator is constructed for a reference ultrasound transducer and expanded into a general measurement geometry using transformations that map the formulated forward operator in local coordinates to the global coordinates of the measurement geometry. The inverse problem is solved using a Bayesian framework. The approach is evaluated with numerical simulations and experimental data. The results show that the proposed approach produces accurate 3-D photoacoustic images with a significantly reduced computational cost both in memory requirements and time. In the studied cases, depending on the computational factors, such as discretization, over the 30-fold reduction in memory consumption was achieved without a reduction in image quality compared to a conventional approach.

Efficient emission modelling in lean premixed flames with pre-tabulated formation characteristics

It is practically important to accurately predict NOx and CO emissions in lean premixed flames for the development of fuel-efficient and low-emission combustion systems. In this study, the efficient modelling with pre-tabulated formation characteristics of NOx and CO is systemically investigated. Their formation characteristics for turbulent flames in the flamelet and thin reaction zone regimes are first quantified with one-dimensional adiabatic premixed flames through the incorporation of turbulence induced diffusion. Results show that the formation of NOx and CO can be characterized by two different stages, i.e., a rapid increase of both in flame front and the linear growth for NOx and exponential decay for CO in post-flame zone. An efficient NOx and CO modelling approach is formulated and demonstrated in a full-scale methane gas turbine combustor, in which species NOx and CO are transported and solved, with the source terms being modelled by pre-tabulated formation characteristics from one-dimensional premixed flames with detailed chemical kinetics. Realizable k-epsilon model and finite rate/eddy dissipation model in conjunction with a two-step global mechanism are employed to primarily predict the flow and flame characteristics. Results show that the predicted NOx and CO agree with experimental measurements over a wide range of equivalence ratios. The discrepancy in NOx emission can be accounted by fuel/air unmixedness. For CO, the rapid increase due to the incomplete CO oxidation resulting from its increasing characteristic oxidation time near lean blow-out (LBO) is correctly captured. It is further shown that the predicted CO emission is sensitive to combustion/kinetics model parameters and the accurate prediction of flame shape and dynamics is crucial to capture the trend of CO formation near LBO.

A simulation-optimization study of the inventory of a bike-sharing system: The case of Mexico City Ecobici’s system

Bike-sharing systems (BSS) are a healthy and environmentally friendly alternative for public transportation in many large urban areas around the world. In 2010, Mexico City started its BSS, named Ecobici, with 84 stations and reached more than 840,000 trips that year. Ecobici rapidly became a preferred option for many users. By 2019, the number of stations in Ecobici had grown almost 6 times since the first year. In addition, the number of registered trips on 2019 was more than 8.4 million and the number of trips in a weekday were more than 30,000. In order to incentivize the adoption of bike as transportation mode, Ecobici and other BSS need to handle the bike inventory at their stations throughout the day. There are many challenges for optimizing bike inventory in BSS given the complex interdependencies, the non-homogeneous demand flows, and the large-scale scenarios. In this paper, we propose and study a simulation–optimization framework to minimize the number of shortage events (failed bike withdrawals and returns) by setting an adequate bike inventory at each station throughout the day. We propose an efficient modeling approach to simulate large-scale BSS and a simulation–optimization heuristic to optimize bike inventory. The experiments show that these algorithms obtain results with similar quality than other search methods, but in a significant shorter amount of time.

An Enhanced Equivalent Circuit Model With Real-Time Parameter Identification for Battery State-of-Charge Estimation

This article introduces an efficient modeling approach based on the Wiener structure to reinforce the capacity of classical equivalent circuit models (ECMs) in capturing the nonlinearities of lithium-ion (Li-ion) batteries. The proposed block-oriented modeling architecture is composed of a simple linear ECM followed by a static output nonlinearity block, which helps achieving a superior nonlinear mapping property while maintaining the real-time efficiency. The observability of the established battery model is analytically proven. This article also introduces an efficient parameter estimator based on extended-kernel iterative recursive least squares algorithm for real-time estimation of the parameters of the proposed Wiener model. The proposed approach is applied for state-of-charge (SoC) estimation of 3.4-Ah 3.6-V nickel-manganese-cobalt-based Li-ion cells using the extended Kalman filter (EKF). The results show about 1.5% improvement in SoC estimation accuracy compared with the EKF algorithm based on the second-order ECM. A series of real-time tests are also carried out to demonstrate the computational efficiency of the proposed method.

Efficient mixed model approach for large-scale genome-wide association studies of ordinal categorical phenotypes

In genome-wide association studies, ordinal categorical phenotypes are widely used to measure human behaviors, satisfaction, and preferences. However, because of the lack of analysis tools, methods designed for binary or quantitative traits are commonly used inappropriately to analyze categorical phenotypes. To accurately model the dependence of an ordinal categorical phenotype on covariates, we propose an efficient mixed model association test, proportional odds logistic mixed model (POLMM). POLMM is computationally efficient to analyze large datasets with hundreds of thousands of samples, can control type I error rates at a stringent significance level regardless of the phenotypic distribution, and is more powerful than alternative methods. In contrast, the standard linear mixed model approaches cannot control type I error rates for rare variants when the phenotypic distribution is unbalanced, although they performed well when testing common variants. We applied POLMM to 258 ordinal categorical phenotypes on array genotypes and imputed samples from 408,961 individuals in UK Biobank. In total, we identified 5,885 genome-wide significant variants, of which, 424 variants (7.2%) are rare variants with MAF < 0.01.

A hybrid shallow water solver for overland flow modelling in rural and urban areas

This study aims to present an accurate and efficient finite volume approach for overland flow modelling in rural and urban areas through solving the two-dimensional shallow water equations. The proposed approach introduces a hybrid procedure that utilizes the sharpness of discontinuity to identify flow conditions in the entire computational domain, and then a low-accuracy/high-efficiency scheme (the Rusanov scheme) is used in regular flows, whereas a high-accuracy/low-efficiency scheme (the HLLC scheme) is adopted in strong discontinuous flows. Regions that simultaneously adopt the two schemes are artificially specified to diminish numerical unbalances between the two schemes. Model calibration is carried out through three strong discontinuous flow cases with the exact solutions. Model verification is next described by three benchmark cases, including rainfall-runoff around buildings, flash floods towards buildings, and street junction flows. Model efficiency assessment is also conducted to present the numerical efficiency improvement of the proposed approach for various flow conditions. Technical attention is devoted to the numerical performances on local accuracy and global efficiency of the proposed approach. The simulated results indicate that the proposed approach can simulate as accurate as the HLLC scheme with significant reduction on its computational time. How efficient the proposed hybrid approach can reach depends on flow conditions involved. It is more efficient than the HLLC scheme as the portion of strong discontinuous flows in the computational domain is relatively smaller. The proposed approach has proved its accuracy and efficiency for overland flow modeling, and it can also be used together with other speed-up techniques, such as local time stepping and parallelization, to improve its efficiency. Other finite volume schemes can also be utilized to improve the efficiency and accuracy of the proposed approach. Therefore, the solver has considerable potentials as a useful tool in flood inundation simulations.

Void-dynamics in nano-wires and the role of microstructure investigated via a multi-scale physics-based model

With scaling of nano-interconnect linewidth toward a 3 nm technology node, electromigration in copper nano-interconnects is becoming a major limitation. In this context, the increase of texture polycrystallinity plays a major role and necessitates better understanding of the impact of void dynamics and its interaction with texture at operating conditions. To this end, comprehensive, yet efficient, physics-based numerical models are warranted given that electromigration experiments at low operating temperatures are not feasible. Albeit, development of such models has been a challenge since it involves large length-scale discrepancy between features, i.e., from wire dimensions (hundreds of micrometers) down to grains and voids (tens of nano-meters and below) leading to substantial computational cost. To this end, in this study an efficient multi-scale physics-based electromigration modeling approach is demonstrated, where an experimentally calibrated Korhonen-type 1D model solves electromigration at the global scale (entire interconnect), and a 2D local model simulates void dynamics considering the impact of electron wind, void surface energy, and stress gradients. The role of copper texture, i.e., grain boundaries, grain orientation and anisotropic properties is thoroughly investigated. We demonstrate that by tailoring the copper grain structure, void behavior and thus resistance evolution of nanowires can be effectively controlled. Compared to wider interconnects, in scaled nanowires, surface energy is found to play a dominant role on void morphology compared to electron wind. Strong anisotropy of diffusivity of copper's FCC lattice fosters faceted morphologies and emergence of slit morphology during void transition through the polycrystalline texture. Bamboo segments are shown to effectively pin the voids migrating toward the cathode in cases with cobalt cap. In addition, voids migrating from low diffusivity to high diffusivity regions adopt a longitudinally elongated morphology, which can be detrimental in a down-stream operation mode.

Analysis of the effects of microstructure heterogeneity on the mechanical behavior of additively manufactured Ti6Al4V using mechanics of structure genome

In this paper, the mechanics of structure genome (MSG) is introduced as a computationally efficient multiscale modeling approach to predict the effects of heterogeneous microstructure on the mechanical properties of AMed Ti6Al4V. Homogenization using MSG is conducted on a structure genome (SG) that incorporates the basic building details of the heterogeneous microstructure from the grain level and the modeling is carried out using the constitutive behaviors of the constituent phases with the evolution of microstructure without redoing homogenization. The microstructure-based MSG was employed to simulate the compression and tensile behaviors of AMed Ti6Al4V based on their microstructures. The predicted true stress-strain curves were then compared to the corresponding curves acquired from experimental testing. The simulation results matched well with the experimental results and showed the existence of martensite to be the reason behind the high strength in the AMed Ti6Al4V and the reduced ductility. Additionally, the inclusion of the microstructure in the SG allowed the MSG simulations to conform to the experiments in showing the reductions in the strengths and improvements in the maximum elongations in the AMed+heat treated Ti6Al4V due to grain coarsening and absence of martensite.

New fundamental period formulae for soil-reinforced concrete structures interaction using machine learning algorithms and ANNs

The importance of designing safe and economic structures in seismically active areas is of great importance. Thus, developing tools that would help in accurately predicting the dynamic properties of buildings is undoubtable a crucial objective. One of the parameters that significantly affects the seismic design of any structure is the fundamental period that is used to compute the seismic forces. It is well documented that the current design formulae for the prediction of the fundamental period of reinforced concrete buildings are simplistic and often fail to capture accurately their expected natural frequency. In addition, the design formulae do not have the ability to account for the soil-structure interaction (SSI) effect that, in some cases, significantly affects the natural frequency of buildings due to the additional flexibility induced by the soft soil. In this research work, a computationally efficient and robust 3D modeling approach is used for the modal analysis in order to investigate the accuracy of different design formulae in predicting the fundamental period of reinforced concrete buildings with and without SSI effects. In this context, 3D detailed modeling is used to generate a dataset that consists of 475 modal analyses, which is subsequently used to train and produce three predictive formulae using a higher-order, nonlinear regression modeling framework. The developed fundamental period formulae were validated through the use of 60 out-of-sample modal results and they are also compared to other existing formulae in the international literature and design codes. According to the numerical findings, the proposed fundamental period formulae are found to have superior predictive capabilities for the under-study types of buildings.

A fast computational model for near- and far-field noise prediction due to offshore pile driving.

This paper presents a computationally efficient modeling approach for predicting underwater noise radiation from offshore pile driving. The complete noise prediction model comprises two modules. First, a sound generation module is adopted to capture the interaction between the pile, the fluid, and the seabed, aiming at modeling the sound generation and propagation in the vicinity of the pile. Second, a sound propagation module is developed to propagate the sound field at larger distances from the pile. To couple the input wavefield obtained from the sound generation module, the boundary integral equations (BIEs) are formulated based on the acousto-elastodynamic reciprocity theorem. To advance the mathematical formulation of the BIEs, the Green's tensor for an axisymmetric ring load is derived using the complex wavenumber integration technique. The model advances the computational efficiency and flexibility of the noise prediction in both near- and far-fields from the pile. Finally, model predictions are benchmarked against a theoretical scenario and validated using measurement data from a recent offshore pile-installation campaign.

HOLISMOKES

Modeling the mass distributions of strong gravitational lenses is often necessary in order to use them as astrophysical and cosmological probes. With the large number of lens systems (≳105) expected from upcoming surveys, it is timely to explore efficient modeling approaches beyond traditional Markov chain Monte Carlo techniques that are time consuming. We train a convolutional neural network (CNN) on images of galaxy-scale lens systems to predict the five parameters of the singular isothermal ellipsoid (SIE) mass model (lens center x and y, complex ellipticity ex and ey, and Einstein radius θE). To train the network we simulate images based on real observations from the Hyper Suprime-Cam Survey for the lens galaxies and from the Hubble Ultra Deep Field as lensed galaxies. We tested different network architectures and the effect of different data sets, such as using only double or quad systems defined based on the source center and using different input distributions of θE. We find that the CNN performs well, and with the network trained on both doubles and quads with a uniform distribution of θE &gt; 0.5″ we obtain the following median values with 1σ scatter: Δx = (0.00−0.30+0.30)″, Δy = (0.00−0.29+0.30)″, ΔθE = (0.07−0.12+0.29)″, Δex = −0.01−0.09+0.08, and Δey = 0.00−0.09+0.08. The bias in θE is driven by systems with small θE. Therefore, when we further predict the multiple lensed image positions and time-delays based on the network output, we apply the network to the sample limited to θE &gt; 0.8″. In this case the offset between the predicted and input lensed image positions is (0.00−0.29+0.29)″ and (0.00−0.31+0.32)″ for the x and y coordinates, respectively. For the fractional difference between the predicted and true time-delay, we obtain 0.04−0.05+0.27. Our CNN model is able to predict the SIE parameter values in fractions of a second on a single CPU, and with the output we can predict the image positions and time-delays in an automated way, such that we are able to process efficiently the huge amount of expected galaxy-scale lens detections in the near future.

Modelling the impact of weather parameters on the microbial quality of water in distribution systems

In this study, a framework for integrating weather variables and seasons into the modelling and prediction of the microbial quality in drinking water distribution networks is presented. Statistical analysis and Bayesian network (BN) modelling were used to evaluate relationships among water quality parameters in distribution pipes and their dependencies on weather parameters. Two robust predictive models for Total Bacteria in the network were built based on a deep learning approach (Long Short-Term Memory (LSTM)). The first model included water quality parameters alone as inputs while the second model included weather parameters. The seven-year dataset used in this study constituted water quality parameters measured at seven location in the water distribution network for the city of Ålesund in Norway, and weather data for the same period. Results of the initial statistical analysis and the BN models showed that, air temperature, the summer season, precipitation, as well as water quality parameters namely, residual chlorine, water temperature, alkalinity and electrical conductivity have strong relations with the counts of Total Bacteria in the distribution networks studied. It was found that the integration of the weather parameters in the Total Bacteria prediction models significantly improved the quality of the predictions. Compared to the LSTM 1, LSTM 2 achieved MAE and MSE values as high as to 6.8 and 4.9 times respectively when the model was tested on the seven locations. In addition, the R2 values were marginally higher in LSTM 2 (0.92–0.95) than in LSTM (0.81–0.86). The prediction results demonstrate the relevance of integrating weather parameters such as air temperature seasons in predicting bacteria levels in water distribution systems. This suggests that changes in the microbial quality of water in distribution systems and potentially drinking water sources could be reliably assessed by integrating online sensors of water quality and weather parameters with efficient models such as the LSTM. Applying this efficient modelling approach in the management of water supply systems could offer immense support in addressing current challenges in assessing the microbial quality of water and minimizing associated health risks.

Genetic algorithm initialized artificial neural network based temperature dependent small‐signal modeling technique for GaN high electron mobility transistors

This paper explores and develops efficient temperature-dependent small-signal modeling approaches for GaN high electron mobility transistors (HEMTs). The multilayer perceptron (MLP) architecture and cascaded MLP architecture of artificial neural network are employed to model temperature dependence of 2-mm GaN-on-silicon device. It is identified that both architectures face problem of dependence on initials values of weights and biases. To overcome this issue, the genetic algorithm (GA) is incorporated in both MLP and cascaded MLP architectures. The models are trained on a large set of operating conditions (bias voltages and ambient temperatures) over a frequency range of 0.1 to 20 GHz and then tested for both temperature interpolation and extrapolation cases to assess their accuracy and robustness. An excellent agreement between the measured and the modeled S-parameters over the entire frequency range demonstrate the quality and robustness of the proposed technique. It is also shown that the cascaded MLP with GA exhibits better performance but with increased complexity.