High Computational Efficiency Research Articles

Stability Analysis in Milling Based on the Localized Differential Quadrature Method.

Chatter stability analysis is an effective way to optimize the cutting parameters and achieve chatter-free machining. This paper proposes a milling chatter stability analysis method based on the localized differential quadrature method (LDQM), which has the advantages of simple principle, easy application, and high computational efficiency. The milling process, considering the regeneration effect, is modeled using linear periodic delay differential equations (DDE), then the state transition matrix during the adjacent cutting period is constructed based on the LDQM, and finally, the stability of the milling process is obtained based on the Floquet theory. The accuracy and computation efficiency of the proposed method are verified through two benchmark milling models. The simulation results demonstrate that the proposed method in this paper can accurately and quickly obtain the chatter stability lobe diagram (SLD) of the milling process, which will provide guidance for optimizing the process parameters.

A unified core model of double-gate and surrounding-gate MOSFETs for circuit simulation

This paper presents a new core compact model of double-gate (DGFET) and surrounding-gate (SGFET) MOSFETs for circuit simulations. The current and the terminal charges are continuous with high computation efficiency and accuracy. Despite its accuracy, it retains the same simplicity of the industry standard transistors models. The drain current is worked out without invoking the charge-sheet approximation exploiting a quadratic symmetric polynomial interpolation of the charge in the channel. Apart this clear approximation, no other simplification is used to work out the drain current, the terminal charges, the potential, and electric field in the channel. The accuracy of the model is shown by comparison with the exact numerical solution and experimental data of the literature.

Multi-Label Weighted Contrastive Cross-Modal Hashing

Due to the low storage cost and high computation efficiency of hashing, cross-modal hashing has been attracting widespread attention in recent years. In this paper, we investigate how supervised cross-modal hashing (CMH) benefits from multi-label and contrastive learning (CL) by overcoming the following two challenges: (i) how to combine multi-label and supervised contrastive learning to consider diverse relationships among cross-modal instances, and (ii) how to reduce the sparsity of multi-label representation so as to improve the similarity measurement accuracy. To this end, we propose a novel cross-modal hashing framework, dubbed Multi-Label Weighted Contrastive Hashing (MLWCH). This framework involves compact consistent similarity representation, a new designed multi-label similarity calculation method that efficiently reduces the sparsity of multi-label by reducing redundant zero elements. Furthermore, a novel multi-label weighted contrastive learning strategy is developed to significantly improve hashing learning by assigning similarity weight to positive samples under both linear and non-linear similarities. Extensive experiments and ablation analysis over three benchmark datasets validate the superiority of our MLWCH method, especially over several outstanding baselines.

A study on predicting students’ grades for ideological and political courses with decision tree generation rules

Predicting students’ course grades is an essential element in teaching. This paper used decision tree generation rules to study the prediction of students’ ideological and political course grades. Firstly, ID3 and C4.5 algorithms were briefly introduced; then, an improved C4.5 algorithm with higher computational efficiency was put forward. The formula of the C4.5 algorithm was optimized using theories such as the Taylor series. Finally, experiments were performed on the UCI dataset and students’ ideological and political course datasets. The results suggested that the average classification accuracy and computation time of the improved C4.5 algorithm was 79.37% and 74.1 ms, respectively, on the UCI dataset, which was better than the traditional C4.5 algorithm. Then, the experiment predicting students’ course grades demonstrated that the average quiz grade and the number of video views had the greatest impact on the final grades. The prediction accuracy of the improved C4.5 algorithm reached 93.46%, and the average computation time was 54.8 ms, which was 19.17% less than the C4.5 algorithm. The experimental results verify the effectiveness of the generation rule of the improved C4.5 algorithm in predicting students’ ideological and political course grades. This algorithm can be applied in the actual grade prediction.

DEEP LEARNING-BASED STEREO MATCHING FOR HIGH-RESOLUTION SATELLITE IMAGES: A COMPARATIVE EVALUATION

Abstract. Dense matching plays an important role in 3D modeling from satellite images. Its purpose is to establish pixel-by-pixel correspondences between two stereo images. The most well-known algorithm is the semi-global matching (SGM), which can generate high-quality 3D models with high computational efficiency. Due to the complex coverage and imaging condition, SGM cannot cope with these situation well. In recent years, deep learning-based stereo matching has attracted wide attention and shown overwhelming benefits over traditional algorithms in terms of precision and completeness. However, existing models are usually evaluated by using close-ranging datasets. Thus, this study investigates the recent deep learning models and evaluate their performance on both close-ranging and satellite image datasets. The results demonstrate that deep learning network can better adapt to the satellite dataset than the typical SGM. Meanwhile, the generalization ability of deep learning-based models is still low for the real application at recent time.

A two-stage model updating method for the linear parts of structures with local nonlinearities

Finite element model updating provides an important supplement for finite element modelling. However, some studies have shown that if the tested structure involves local nonlinearities due to damages, material properties and large deformation et al., it is difficult to achieve an accurate modified model using conventional model updating methods that are based on the assumption of linear structures. To address this issue, a two-stage model updating method separating the effects of local nonlinearities is proposed in this paper. Firstly, the underlying linear frequency response function is obtained by using the conditioned reverse path method. Then, combined with the Sherman-Morrison-Woodbury formula and the model updating objective function established by the frequency response function similarity metric, then structural model updating and damage detection are carried out as the second stage. Three numerical examples are given to illustrate the effectiveness of the proposed method. This method can not only accurately identify the location and quantify the extent of structural damages, but also has the advantages of not based on sensitivity, not depending on the selection of frequency points, not repeatedly calling the initial model et al. The proposed method has high computational efficiency and avoids the numerical problems often encountered by conventional frequency response function-based model updating methods.

Stochastic analysis of thin beams and plates incorporating von-Kármán nonlinear strains using meshless method

The current paper proposes methods for stochastic meshless analysis of thin beams and plates, by considering von-Kármán nonlinear strains. Here, Young’s modulus is assumed to have a spatial variation over the domain and is modeled as a homogeneous random field. Further, it is discretized utilizing moving least square based shape functions. Gaussian and lognormal properties are presumed to model the randomness related to Young’s modulus. Element-free Galerkin method is used as the meshless tool. The present study suggests employing perturbation method if the quantities of interest are limited to the first or second probabilistic moments of the responses. The nonlinear equations in the proposed perturbation formulations are solved using Newton-Raphson iterative scheme for finding out the deterministic part of the response, as well as its initial two derivatives with respect to random variables. Additionally, if there is an interest in higher probabilistic moments, probability density functions, cumulative distribution functions, and so forth, the study proposes a method using high-dimensional model representation (HDMR) in the meshless framework. In HDMR, the nonlinear connection between input variables and the output response is represented in terms of hierarchical correlated function expansions. These hierarchical functions are approximated over the response values evaluated using deterministic analysis on selected number and combinations of control points of random variables involved. Lagrange interpolation method is employed for the construction of response surfaces. Unlike Monte Carlo simulation (MCS) which needs numerous deterministic nonlinear analyses to be performed during each simulation, HDMR requires only a few deterministic nonlinear analyses to construct the component functions. Hence, simulations performed on the reduced order models of high-dimensional response can be used to estimate the probabilistic moments as well as density functions and cumulative distributions of response, with high computational efficiency compared to MCS. The probabilistic moments and distributions produced for a range of coefficient of variation of input random field up to 30%, for both the normal and lognormal input random fields are compared with those provided by crude MCS on the system of equations and are observed to be matching well.

Using deep learning to accurately detect sow vulva size in a group pen with a single camera.

This paper presents a non-contact method for the detection of changes in sow vulva size in a group pen. The traditional approach to estrus detection is manually pressing down on the back of the sow to elicit standing responses; however, this method causes undue distress for sows not in estrus. When a sow is in estrus, the vulva is red and swollen due to the presence of endocrine. Monitoring changes in vulva size to detect estrus with as little impact on the sow as possible is the focus of this study. This is achieved using a single camera combined with a deep learning framework. Our approach comprises two steps: vulva detection and vulva size conversion. Images of sows of Yorkshire, Landrace, and Duroc breeds were collected in group housing, and the vulva was detected through artificial markers and the network architecture of YOLO v4. Based on the internal and external parameters of the camera, the detected size was converted into millimeters and the results of manual measurement (MM) and automatic calculation combined to calculate the size of the vulva. Analysis of the calculated size compared with MM indicates that the object recognition rate of the system exceeds 97.06%, with a size error of only + 1.70 to -4.47mm and high-calculation efficiency (>2.8 frames/s). Directions for future research include the automatic detection of pig width.

Comparative analysis of finite‐difference and split‐step based parabolic equation methods for tunnel propagation modelling

AbstractRadio wave propagation modelling in railway environments is of fundamental importance in designing reliable train communication systems. Parabolic equation (PE) methods have been widely applied to the modelling of wave propagation in tunnels due to their high computational efficiency and fidelity. The finite‐difference parabolic equation (FDPE) and the split‐step parabolic equation (SSPE) methods are two commonly used approaches to solve PE numerically. However, the relevant literature is still missing a comprehensive study of their performance, including the selection of parameters such as discretisation steps and the tradeoffs involved in terms of their accuracy and efficiency, especially as current wireless systems shift to high frequencies. In this study, a systematic analysis of the error and computational complexity of the FDPE and SSPE methods for radio wave propagation modelling in tunnels is provided. Guidelines for the choice of their parameters are provided, and their performance is demonstrated through both numerical examples and experimental measurements in actual tunnel cases.

Improvement of spatial-temporal urban heat island study based on local climate zone framework: A case study of Hangzhou, China

Large-scale long-period urban heat island (UHI) intensity (UHII) prediction with high spatiotemporal resolutions, satisfactory accuracy, and calculation efficiency is crucial and challenging for UHI mitigation studies. This study proposes a framework combining an urban weather generator (UWG), local climate zone (LCZ), deep-learning method, and Python automatic cyclical calculation to obtain hourly UHII throughout one year and over a total of 1920 blocks in Hangzhou City. The spatial-averaged hourly UHII are between −2 °C and 6 °C in more than 96 % time, and those at 0 °C–1 °C contribute to 43.80 % of the total time. Spring gives the most intense nocturnal UHII, while winter has the weakest one. A significant diurnal UCI phenomenon could accompany the strong nocturnal UHI phenomenon. From the synthetic (based on seasonal data) 24-h curves, mean UHII and UCII (average of the positive and negative values, respectively) drop by approximately 25 % in winter compared to spring. Spatially, UHII is higher in the central regions with compact buildings but significantly lower in the high-vegetation-coverage regions. Based on LCZ framework, LCZ 1 (compact high-rise configurations) has the highest UHII, independent of examined periods. UHII relations to building coverage, vegetation ratio, and building height are individually quantified. The building coverage has the highest influence on annual UHII, with a correlation coefficient as high as 0.76. Results indicate that, for UHI mitigation purposes, the percentage of compact high-rise configurations (LCZ 1) in the urban area shall be limited, and the vegetation ratio is better to be greater than 20 %.

Sparse time-frequency analysis of seismic data via convolutional neural network

Time-frequency (TF) analysis is commonly used to reveal the local properties of seismic signals, such as frequency and spectral contents varying with time/depth. Aiming to realize a highly localized TF representation of seismic signals, researchers treated the TF analysis as an inverse problem, and the regularization is adopted in the objective functions. Traditionally, the TF sparse inversion process is solved by the Lasso regression. It has been proven that the Lasso regression needs a large number of iterations to reach a high accurate solution for the convex problem. Recently, convolutional neural network (CNN) has been successfully used to solve the convex problem due to their high computational efficiency and strong nonlinear characterization ability. We use CNN to solve the sparse TF inversion, and our method is called STFA-CNN. The objective function in the neural network architecture consists of two portions. The first one is to minimize the difference between the local forward and backward Fourier transform of seismic signals. The second is minimizing the regularization ([Formula: see text] norm) of TF results. To demonstrate the effectiveness of our method, we apply it to synthetic and real seismic data. We further use the calculated TF spectra to compute the attenuation of seismic waveforms and apply the attenuation attribute to predict the hydrocarbons of a seismic survey acquired over the Ordos Basin, northwest of China.

Identification of two-phase flow patterns in Z-shaped offshore pipelines based on deep learning technologies

Flow pattern identification is of great significance for the flow assurance of multiphase flow in offshore pipelines. For this purpose, a deep learning algorithm integrating convolutional neural networks (CNNs) and long short-term memory (LSTM) networks is developed in this study to identify different flow patterns of two-phase flow in a Z-shaped pipeline. To verify the applicability and performance of the proposed algorithm, a series of experiments are conducted in the multiphase flow loop. Then, the flow pattern pictures, mass flow rates, and vibration signals are collected by the NI data acquisition system and subjected to denoising using the density-based spatial clustering of applications with noise (DBSCAN) algorithm combined with LSTM. Meanwhile, the Kendall rank correlation is employed for feature processing. Finally, the proposed deep learning algorithm achieves accuracies of 0.98077 with engineering parameters and 0.955 with time-domain parameters. In addition, a comparison analysis is conducted with four other common deep learning algorithms in terms of computation time, learning rate, accuracy, and different sample quantities. The results demonstrate that the proposed algorithm can not only improve the identification accuracy but can also achieve faster convergence speed and higher computational efficiency.

Direct Numerical Simulation of Transitional and Turbulent Flows Over Multi-Scale Surface Roughness—Part I: Methodology and Challenges

Abstract High-fidelity simulation of transitional and turbulent flows over multi-scale surface roughness presents several challenges. For instance, the complex and irregular geometrical nature of surface roughness makes it impractical to employ conforming structured grids, commonly adopted in large-scale numerical simulations due to their high computational efficiency. One possible solution to overcome this problem is offered by immersed boundary methods, which allow wall boundary conditions to be enforced on grids that do not conform to the geometry of the solid boundary. To this end, a three-dimensional, second-order accurate boundary data immersion method (BDIM) is adopted. A novel mapping algorithm that can be applied to general three-dimensional surfaces is presented, together with a newly developed data-capturing methodology to extract and analyze on-surface flow quantities of interest. A rigorous procedure to compute gradient quantities such as the wall shear stress and the heat flux on complex non-conforming geometries is also introduced. The new framework is validated by performing a direct numerical simulation (DNS) of fully developed turbulent channel flow over sinusoidal egg-carton roughness in a minimal-span domain. For this canonical case, the averaged streamwise velocity profiles are compared against results from the literature obtained with a body-fitted grid. General guidelines on the BDIM resolution requirements for multi-scale roughness simulation are given. Momentum and energy balance methods are used to validate the calculation of the overall skin friction and heat transfer at the wall. The BDIM is then employed to investigate the effect of irregular homogeneous surface roughness on the performance of an LS-89 high-pressure turbine blade at engine-relevant conditions using DNS. This is the first application of the BDIM to realize multi-scale roughness for transitional flow in transonic conditions in the context of high-pressure turbines. The methodology adopted to generate the desired roughness distribution and to apply it to the reference blade geometry is introduced. The results are compared to the case of an equivalent smooth blade.

A novel semi‐resolved CFD‐DEM method with two‐grid mapping: Methodology and verification

AbstractThe semi‐resolved Computational Fluid Dynamics coupled with the Discrete Element Method (CFD‐DEM) method has emerged as approach to modeling particle‐fluid interactions in granular materials with high particle size ratios. However, challenges arise from conflicting requirements regarding the CFD grid size, which must adequately resolve fluid flow in the pore space while maintaining a physically meaningful porosity field. This study addresses these challenges by introducing a two‐grid mapping approach. Initially, the porosity field associated with fine particles is estimated using a coarse CFD grid, which is then mapped to a dynamically refined grid. To ensure conservation of total solid volume, a volume compensation procedure is implemented. The proposed method has been rigorously verified using benchmark cases, showing its high computational efficiency and accurate handling of complex porosity calculations near the surface of coarse particles. Moreover, the previously unreported impact of the empirical drag correlation on fluid‐particle force calculations for both coarse and fine particles has been revealed.

Improvement of 2D cine image quality using 3D priors and cycle generative adversarial network for low field MRI-guided radiation therapy.

Cine magnetic resonance (MR) images have been used for real-time MR guided radiation therapy (MRgRT). However, the onboard MR systems with low-field strength face the problem of limited image quality. To improve the quality of cine MR images in MRgRT using prior image information provided by the patient planning and positioning MR images. This study employed MR images from 18 pancreatic cancer patients who received MR-guided stereotactic body radiation therapy. Planning 3D MR images were acquired during the patient simulation, and positioning 3D MR images and 2D sagittal cine MR images were acquired before and during the beam delivery, respectively. A deep learning-based framework consisting of two cycle generative adversarial networks (CycleGAN), Denoising CycleGAN and Enhancement CycleGAN, was developed to establish the mapping between the 3D and 2D MR images. The Denoising CycleGAN was trained to first denoise the cine images using the time domain cine image series, and the Enhancement CycleGAN was trained to enhance the spatial resolution and contrast by taking advantage of the prior image information from the planning and positioning images. The denoising performance was assessed by signal-to-noise ratio (SNR), structural similarity index measure, peak SNR, blind/reference-less image spatial quality evaluator (BRISQUE), natural image quality evaluator, and perception-based image quality evaluator scores. The quality enhancement performance was assessed by the BRISQUE and physician visual scores. In addition, the target contouring was evaluated on the original and processed images. Significant differences were found for all evaluation metrics after Denoising CycleGAN processing. The BRISQUE and visual scores were also significantly improved after sequential Denoising and Enhancement CycleGAN processing. In target contouring evaluation, Dice similarity coefficient, centroid distance, Hausdorff distance, and average surface distance values were significantly improved on the enhanced images. The whole processing time was within 20ms for a typical input image size of 512 × 512. Taking advantage of the prior high-quality positioning and planning MR images, the deep learning-based framework enhanced the cine MR image quality significantly, leading to improved accuracy in automatic target contouring. With the merits of both high computational efficiency and considerable image quality enhancement, the proposed method may hold important clinical implication for real-time MRgRT.

Predicting Potential Risk: Cerebral Stroke via Regret Minimization

Objective. The data processing of medical test report has always been one of the important contents in biological information domain, especially the process of extracting the effective information from the report so as to assist doctors with the correct medical plan. Usual methods neglect the implicit relationship between features. More features are generally not a better choice because more noise is generated between feature combinations. We propose a practical feature selection strategy RMFS, which aims to select the optimal combination of features. Materials and Methods. Based on the above situation, in this paper, 64 features are extracted from a real medical test report dataset for stroke and feature selection is defined as a reinforcement learning problem to optimize the feature combination by minimizing regret. We select three current mainstream feature selection methods and conduct comparative experiments. Results. We processed and completed a dataset derived from real medical test reports of stroke. We redefine the feature selection problem as a reinforcement learning problem and propose an optimization strategy based on regret minimization and train weight parameters in a DQN network. Experimental results demonstrate that our method can identify feature combinations with higher prediction accuracy. Discussion. RMFS shows a strong robustness to the randomness of the environment and has high computational efficiency and accuracy. Compared with the previous feature selection methods, our method yields superior results. Conclusion. The experimental results demonstrate that our method can obtain a more accurate prediction rate under the same feature scale and we can achieve baseline performance with fewer features.

A Brain-Inspired Hierarchical Interactive In-Memory Computing System and Its Application in Video Sentiment Analysis

Video sentiment analysis can effectively establish the relationship between the emotion state and the multimodal information, while still suffer from intensive computation and low efficiency, due to the von Neumann computing architecture. Here, we present a brain-inspired hierarchical interactive in-memory computing (IMC) system, which can efficiently solve ‘von Neumann bottleneck’, enabling cross-modal interactions and semantic gap elimination. First, a 1T1M synapse array is fabricated using cost-effective, highly stable, flexible, and eco-friendly carbon materials, offering efficient analog multiply-accumulate operations. To illustrate the complexity of the proposed brain-inspired hierarchical interactive IMC system, three modules are proposed: 1) unimodal extraction module, 2) hierarchical interactive module, 3) output module. Furthermore, the proposed system is validated by applying it to video sentiment analysis. The experimental results demonstrate that the proposed system outperforms the existing state-of-the-art methods with high computational efficiency and good robustness. This work opens up a new way to achieve the deep integration of nanomaterials, deep learning, and modern electronics into IMC.

(Q, S, R)-Dissipativity Analysis of Large-Scale Networked Systems

This brief investigates the (Q,S,R)-dissipativity of large-scale networked systems in the discrete-time case. The networked system consists of a large number of subsystems, each of which can have different dynamic characteristics, and the connections among subsystems are arbitrary. When the number of subsystems is large, the traditional analysis method based on lumped description may have computational difficulties. To solve this problem, some computationally attractive conditions are derived, which essentially depend on the block-diagonal structure of the system parameter matrices and the sparseness of the subsystem connection matrix. Numerical simulations illustrate that compared with the existing results, the conditions suggested in this brief have higher computational efficiency in the dissipative analysis of large-scale networked systems.

A Comparative Analysis of Optimization Algorithms for Finite Element Model Updating on Numerical and Experimental Benchmarks

Finite Element Model Updating (FEMU) is a common approach to model-based Non-Destructive Evaluation (NDE) and Structural Health Monitoring (SHM) of civil structures and infrastructures. Its application can be further utilized to produce effective digital twins of a permanently monitored structure. The FEMU concept, simple yet effective, involves calibrating and/or updating a numerical model based on the recorded dynamic response of the target system. This enables to indirectly estimate its material parameters, thus providing insight into its mass and stiffness distribution. In turn, this can be used to localize structural changes that may be induced by damage occurrence. However, several algorithms exist in the scientific literature for FEMU purposes. This study benchmarks three well-established global optimization techniques—namely, Generalized Pattern Search, Simulated Annealing, and a Genetic Algorithm application—against a proposed Bayesian sampling optimization algorithm. Although Bayesian optimization is a powerful yet efficient global optimization technique, especially suitable for expensive functions, it is seldom applied to model updating problems. The comparison is performed on numerical and experimental datasets based on one metallic truss structure built in the facilities of Cranfield University. The Bayesian sampling procedure showed high computational accuracy and efficiency, with a runtime of approximately half that of the alternative optimization strategies.

Diffusion equation-based modelling of reverberation chambers for sound absorption measurements

Recent research in reverberation chambers has revealed unacceptably large variations resulting from round-robin tests of the standardized measurements of random-incidence absorption coefficients. These variations have prompted active research activities in architectural acoustics. This study applies a diffusion equation model (DEM) in acoustics of reverberation chambers for random incidence absorption measurements by predicting both reverberant sound energies and energy flows. The DEM is of high computational efficiency when compared with other existing geometrical acoustics and wave-based methods. The DEM modeling of the sound energy flows in a reverberation room further reveals violations of the diffuse field assumption near the boundary area of absorbing sample under test. Due to its computational efficiency and higher order of the reverberant energy prediction of the statistical room-acoustic modeling, this work investigates an estimation approach to the absorption coefficients of the sample under test. The proposed approach estimates the absorption coefficients by pursuing both the DEM simulations and the chamber-based measurements. This paper investigates sound energy flux over varied absorbing degrees of absorber samples in standardized sizes, computational efficiency of the DEM for reverberant energy decays in reverberation chambers, and potential significance applied to random-incident absorption coefficient measurements.