High Computational Accuracy Research Articles

Solution of conservative-form transport equations with physics-informed neural network

Physics-informed neural network (PINN) has recently gained significant attention in the field of scientific computing. This paper proposes a PINN framework for solving conservative-form transport equations (PINNforCTE). The PINNforCTE features an accelerated back-propagation in the neural network (NN) training due to low-order derivatives in the governing equations. We demonstrate the application of PINNforCTE in simulating a classical flow and heat transfer problem, i.e., the lid-driven cavity flow and heat transfer under three types of boundary conditions, and the simulation results are compared with those obtained from the traditional PINN and the finite element CFD method. The comparisons indicate that PINNforCTE can significantly accelerate the NN training and potentially shorten the solution time. For all cases considered, the training time of PINNforCTE is reduced by approximately ∼50% compared with the conventional PINN. Furthermore, PINNforCTE can accelerate the Adam optimization algorithm (L-BFGS-B optimization algorithm) by ∼100% (∼80%) compared to the traditional PINN. Additionally, PINNforCTE shows greater convenience in applying the Neumann and Robin boundary conditions directly and higher calculation accuracy than the conventional PINN. Finally, the ability of PINNforCTE to solve the inverse problems is confirmed with an example of 2D steady lid-driven cavity flow and heat transfer. PINNforCTE is an efficient NN solver for conservative-form transport equations (CTEs) that are ubiquitous in science and engineering.

Toward Practical Lightweight Passive Human Tracking Using WiFi Sensing

With the wide adoption of versatile IoT devices, device providers may desire to locate users around those devices to plan context-aware intelligence, which may improve the quality of daily life. As most IoT devices are Wi-Fi enabled, the Wi-Fi-based indoor positioning system is supposed to achieve this future scene. However, the state-of-art Wi-Fi indoor positioning systems face challenges when being practically deployed as they may have to trade-off between, say accuracy and computational overhead. In light of this, this article mainly introduces PLP-Track, a Practical Lightweight Passive indoor tracking system based on CSI (channel state information) fingerprints. To settle the low granularity of fingerprints in distinguishing different positions, we propose a fingerprint preprocessing algorithm based on unsupervised learning and incorporate this algorithm with a state-space model to enable lightweight real-time tracking. Our implementation and evaluation of commodity WiFi devices demonstrate that PLP-Track can achieve indoor localization with high accuracy and low computation cost.

A Low-Power DNN Accelerator Enabled by a Novel Staircase RRAM Array.

Enhancing the ubiquitous sensors and connected devices with computational abilities to realize visions of the Internet of Things (IoT) requires the development of robust, compact, and low-power deep neural network accelerators. Analog in-memory matrix-matrix multiplications enabled by emerging memories can significantly reduce the accelerator energy budget while resulting in compact accelerators. In this article, we design a hardware-aware deep neural network (DNN) accelerator that combines a planar-staircase resistive random access memory (RRAM) array with a variation-tolerant in-memory compute methodology to enhance the peak power efficiency by 5.64× and area efficiency by 4.7× over state-of-the-art DNN accelerators. Pulse application at the bottom electrodes of the staircase array generates a concurrent input shift, which eliminates the input unfolding, and regeneration required for convolution execution within typical crossbar arrays. Our in-memory compute method operates in charge domain and facilitates high-accuracy floating-point computations with low RRAM states, device requirement. This work provides a path toward fast hardware accelerators that use low power and low area.

Study on the influence of chain extension and substituent position on charge transport properties and optoelectronic properties based on 2,5-di(pyrazine-2-yl)furan derivatives

In this paper, the basic structure of 2,5-di(pyrazine-2-yl)furan (DPF) was initially designed to tune the charge transport properties and photoelectric properties of organic materials by chain extension. Through the calculation of recombination energy, the structure with the introduction of anthracene group (DPF-A) is a stable P-type material with the smallest recombination energy in terms of adiabatic ionization potential, so the chain extended molecule DPF-A was selected as the base for the study of substituent positions. While thiophene, trifluorinated acyl, and methoxy groups were introduced to design DPF-A series derivatives. Density functional theory with high computational accuracy was applied to investigate geometric structure, orbital energy levels, ionization energy, electron affinity energy, and recombination energy. The results show that the increase (decrease) of the HOMO (LUMO) energy level leads to the red (blue)-shift of absorption wavelength and fluorescence emission wavelength; the vertical ionization potential decreases and the vertical electron affinity potential increases, thus the hole and electron injection barrier decreases; the decrease of the hole/electron recombination energy improves the charge transport ability of the molecule. Moreover, the adjustment of thiophene orientation and the position of the introduced group has the potential to transform the p-type material into an n-type material, resulting in the enhanced electron injection ability.

An accurate solution method for the dynamic characteristics of three-layered partial-interaction composite beams

In the dynamic analysis of three-layered partial interaction composite beams (PICBs), it is not clear whether the effects of shear deformation and rotary inertia all need to be considered, and the shear-locking problems often occurs in existing finite element methods. To improve computational efficiency, accuracy and stability, an analytical method is proposed by considering the shear slips at the interface and the independent rotary angles among the sub-elements. In the proposed method, a matrix containing the natural frequencies of the three-layered PICB is appropriately established by using the solutions to differential equations of motion rather than the shape function in finite element methods. The main improvement of the proposed method is the analyticity of solutions, which can effectively avoid shear-locking problems often mentioned in finite element methods and obtain accurate solutions due to no introduction of approximate shape functions. The present results agree well with the analytical and numerical results in the literature, which validates the accuracy of the proposed method. The effects of shear deformation and rotary inertia on the dynamic behaviors of three-layered PICBs are discussed. It reveals that the shear deformation can not be ignored but rotary inertia can be ignored in the dynamic analysis of three-layered PICBs, which can effectively reduce the computation complexity. In general, the proposed method has practical value in natural frequency analysis of the three-layered PICB with high computational efficiency, accuracy and stability.

Theoretical analysis and comparative study of top 10 optimization algorithms with DMS algorithm

The significance of big data are prone to complication in solving optimization issues. In several scenarios, one requires adapting several contradictory goals and satisfies various criterions. This made the research on multi-objective optimization more vital and has become main topic. This paper presents theoretical analysis and comparative study of top ten optimization algorithms with respect to DMS. The performance analysis and study of optimization algorithms in big data streaming are explicated. Here, the top ten algorithms of optimization based on recency and popularity are considered. In addition, the performance analysis based on Efficiency, Reliability, Quality of solution, and superiority of DMS algorithm over other top 10 algorithms are examined. From analysis, the DMS provides better efficiency as it endeavours less computational effort to generate better solution, due to acquisition of both DA and MS algorithm’s benefits and DMS takes less time to process a task. Moreover, the DMS needs less number of iterations in the process of optimization and helps to stop optimization process in local optimum. In addition, the DMS has better reliability as it poses the potential to handle specific level of performance. In addition, the DMS utilizes heuristic information for attaining high reliability. Moreover, the DMS produced high computation accuracy, which reveals its solution quality. From the analysis, it is noted that DMS attained improved outcomes in terms of efficiency, reliability and solution quality in contrast to other top 10 optimization algorithms.

Reduced Order Modeling of Dynamic Mechanical Metamaterials for Analysis of Infinite and Finite Systems

Abstract Dynamic mechanical metamaterials (MMs) are artificial media composed of periodic micro-structures, designed to manipulate wave propagation. Modeling and designing MMs can be computationally demanding due to the broad design space spanned by the geometric and material parameters. This work aims to develop a generalized reduced order modeling approach for determining MM dynamics in low frequency ranges with accuracy and speed, using a limited number of parameters and small matrices. The MM unit cells are treated as assemblies of structural elements with discrete degrees-of-freedom, whose effective stiffness and inertia are determined by optimizing energy criteria based on continuum results derived from a small number of eigen-study simulations. This proposed approach offers a parameterized and discretized representation of MM systems, which leads to fast and accurate computation of eigen-study results for periodic arrays, as well as dynamic responses in time domain for finite-sized arrays. The high computational efficiency and physical accuracy of this method will help streamline the modeling process and aid in design discovery and optimization, especially in combination with machine learning and data-driven techniques.

Mechanical behaviour of SPSWs with a central opening reinforced by CFS members

It is often necessary to set up doors, windows, or equipment plumbing openings on the steel plate shear walls (SPSWs) to meet the functional requirements of the building. Meanwhile, to reduce the weakening of the mechanical performance of SPSWs caused by openings, it is usually necessary to weld stiffeners around the openings. However, welded stiffeners often bring residual stress and deformation to the structures. Therefore, the prefabricated stiffeners for opening reinforcement, cold-formed steel (CFS) members, were proposed in this study. To investigate the mechanical performance of SPSWs with a central opening reinforced by CFS members, the quasi-static test was carried out, and the failure mode and hysteresis curve were examined. Besides, compared with the SPSW that was the same size and without openings, the effects of openings on the lateral strength, energy dissipation capacity, and stiffness degradation were analysed. Based on the finite element (FE) model verified by the experiment results, the influences of the arrangements of the CFS members, the opening size, and the width-to-height ratio of the opening were analysed. Finally, according to the reinforcement mechanism of CFS members around openings, a simplified analysis model (SAM) was proposed. In SAM, SPSW with openings reinforced by CFS members can be equivalent to the SPSW without openings, which helps to simplify the calculation greatly. The results showed that the SAM proposed in this study had high calculation accuracy when compared with the experiment results and the FE results.

Implementation of a coupled FEM-SBFEM for soil-structure interaction analysis of large-scale 3D base-isolated nuclear structures

Accurately simulating the radiation damping of an infinite domain is a crucial issue for reasonably and effectively evaluating the isolation effect of three-dimensional (3D) base-isolated nuclear structures considering soil-structure interaction (SSI). The scaled boundary finite element method (SBFEM) has excellent advantages that enable it to precisely fulfil the radiation conditions. In this study, the finite element method (FEM) and SBFEM were combined to develop a coupled FEM-SBFEM technology. The structure and finite domain were discretized using FEM, and the infinite soil was realised using SBFEM. A complete and efficient calculation procedure for SSI analysis was developed based on the coupled FEM-SBFEM, improving the calculation efficiency through parallel programming. The accuracy of the developed procedure was verified using a numerical example. Furthermore, the dynamic response of a large-scale refined 3D base-isolated reactor building considering SSI was analysed using the proposed method. The results demonstrate that the 3D base isolation technology has excellent isolation performance. Additionally, we compared the response results of the proposed coupled FEM-SBFEM and direct method. The proposed procedure has broad engineering applicability and can be used for seismic SSI analysis of large-scale nuclear structures while maintaining high computational accuracy and stability.

Forced vibration analysis of thin cross-ply laminated circular cylindrical shells with arbitrary boundary conditions using the symplectic wave-based method

A symplectic wave-based method is proposed in this paper for the forced vibration analysis of thin cross-ply laminated cylindrical shells with arbitrary boundary conditions. Firstly, based on Kirchhoff–Love’s shell theory, the unified governing equation of thin cross-ply laminated cylindrical shells in the symplectic duality system is established by selecting an appropriate state vector, which leads to the transformation of the vibration analysis of cylindrical shells from two-dimensional to one-dimensional. Then, the response of laminated cylindrical shells can be described in terms of wave shapes in the symplectic duality system. The relationship between incident and reflected waves at a boundary is established to analytically describe the boundary condition, which allows the present method to deal with arbitrary boundary conditions and has high computational accuracy. Finally, the displacement and internal force responses can be simultaneously and explicitly calculated by using the adjoint symplectic orthogonal relation of wave shapes in the symplectic duality system, which ensures that the present method has high computational efficiency. The present method has truncation errors only in the circumferential direction, and is analytical in the axial direction, which can lead to a high accuracy and convergence rate. Several examples demonstrate the effectiveness, convergence and accuracy of the present method, as well as its ability to handle arbitrary boundary conditions.

Use of Composite Multivariate Multiscale Permutation Fuzzy Entropy to Diagnose the Faults of Rolling Bearing.

The study focuses on the fault signals of rolling bearings, which are characterized by nonlinearity, periodic impact, and low signal-to-noise ratio. The advantages of entropy calculation in analyzing time series data were combined with the high calculation accuracy of Multiscale Fuzzy Entropy (MFE) and the strong noise resistance of Multiscale Permutation Entropy (MPE), a multivariate coarse-grained form was introduced, and the coarse-grained process was improved. The Composite Multivariate Multiscale Permutation Fuzzy Entropy (CMvMPFE) method was proposed to solve the problems of low accuracy, large entropy perturbation, and information loss in the calculation process of fault feature parameters. This method extracts the fault characteristics of rolling bearings more comprehensively and accurately. The CMvMPFE method was used to calculate the entropy value of the rolling bearing experimental fault data, and Support Vector Machine (SVM) was used for fault diagnosis analysis. By comparing with MPFE, the Composite Multiscale Permutation Fuzzy Entropy (CMPFE) and the Multivariate Multiscale Permutation Fuzzy Entropy (MvMPFE) methods, the results of the calculations show that the CMvMPFE method can extract rolling bearing fault characteristics more comprehensively and accurately, and it also has good robustness.

Systematic Study of Heteroarene Stacking Using a Congeneric Set of Molecular Glues for Procaspase-6.

Heteroaromatic stacking interactions are important in drug binding, supramolecular chemistry, and materials science, making protein-ligand model systems of these interactions of considerable interest. Here we studied 30 congeneric ligands that each present a distinct heteroarene for stacking between tyrosine residues at the dimer interface of procaspase-6. Complex X-ray crystal structures of 10 analogs showed that stacking geometries were well conserved, while high-accuracy computations showed that heteroarene stacking energy was well correlated with predicted overall ligand binding energies. Empirically determined KD values in this system thus provide a useful measure of heteroarene stacking with tyrosine. Stacking energies are discussed in the context of torsional strain, the number and positioning of heteroatoms, tautomeric state, and coaxial orientation of heteroarene in the stack. Overall, this study provides an extensive data set of empirical and high-level computed binding energies in a versatile new protein-ligand system amenable to studies of other intermolecular interactions.

A cyclic-softening macro element model for mono-bucket foundations supporting offshore wind turbines in clay

A cyclic-softening macro element model that represents mono-bucket foundation supporting offshore wind turbines in clay is presented in this paper. The model is formulated based on multi-surface plasticity with kinematic and isotropic hardening rule. The softening behaviour of the foundation subjected to a cyclic load sequence is captured real-timely in the macro element model through the contraction of yield surfaces. The model parameters are calibrated by uniaxial quasi-static load–displacement curves, which are obtained by finite element analyses with the cyclic behaviour of soil simulated via the cyclic contour diagram approach. The performance of the macro element model in representing the softening of a bucket foundation during a cyclic load sequence is demonstrated by comparing with parallel finite element analyses. A reasonable agreement has been achieved. The procedure to apply the model is suggested, considering different working scenarios with various cyclic to average load ratios. The macro element model developed is capable of capturing the real-time displacement of foundation in a load sequence, with high computational efficiency and accuracy.

Inverse design of 3D cellular materials with physics-guided machine learning

This paper investigates the feasibility of data-driven methods in automating the engineering design process, specifically studying inverse design of cellular mechanical metamaterials. Traditional methods of designing cellular materials typically rely on trial and error or iterative optimization, which often leads to limited productivity and high computational costs. While data-driven approaches have been explored for the inverse design of cellular materials, many of these methods lack robustness and fail to consider the manufacturability of the generated structures. This study aims to develop an efficient inverse design methodology that accurately generates mechanical metamaterial while ensuring the manufacturability of predicted structures. To achieve this, we have created a comprehensive dataset that spans a broad range of mechanical properties by applying rotations to cubic structures synthesized from the nine cubic symmetries of cubic materials. We then employ a physics-guided neural network (PGNN) consisting of dual neural networks: a generator network, which serves as the inverse design tool, and a forward network, which acts as a physics-guided simulator. The goal is to generate desired anisotropic stiffness components with unit-cell design parameters. The results of our inverse model are examined with three distinct datasets and demonstrate high computational efficiency and prediction accuracy compared to conventional methods.

The elastoplastic large deformation analysis based on meshless radial basis reproducing kernel particle method

In this paper, the radial basis reproducing kernel particle method (RB-RKPM) is extended to solve the elastoplastic large deformation problem. The Galerkin weak form of the elastoplastic large deformation problem based on the full Lagrange scheme is used to establish the solution equation, which is then solved using the Newton–Raphson iterative method. The RB-RKPM is established for the elastoplastic large deformation problem. The effect of the penalty factor, the scale parameter, and the shape parameter on the calculation accuracy are introduced, and the optimal parameters are obtained. Several numerical examples were used in the simulation. The RB-RKPM solution is compared with the reproducing kernel particle method (RKPM) solution and the finite element method (FEM) solution, the effectiveness of the RB-RKPM in solving elastoplastic large deformation problem is verified, and the RB-RKPM has high computational accuracy.

Discrete-Time Current Control of Salient Machines With a Simplified Model

The discrete-time model for salient machines with accurate discretization is quite complicated due to different <i>dq</i>-axes inductances. This paper proposes an approximation for the discrete-time model with high accuracy and efficient calculation. Various patterns of pulse-width-modulation updating and current sampling that result in one-sample or a-half-sample computational delay are studied, and corresponding current controllers are directly designed in the discrete-time domain with the simplified model. Decoupled responses between <i>dq</i>-axes are realized in reference tracking. Active resistance is also employed to improve the disturbance rejection. The experimental verification is performed on an interior permanent magnet synchronous machine.

Calculation method of benchmark value of line loss rate in transformer district considering marketing customer portrait

In order to quantitatively analyze the benchmark value of line loss rate in the transformer district, understand the correlation and difference between samples in each transformer district, and improve the dynamic management ability of line loss in the transformer district, a method for calculating the benchmark value of line loss rate in the transformer district based on the portrait of marketing customers is proposed. The high-resolution feature quantity of the data in the transformer district is mined by the feature analysis method of marketing customer portrait, and the electrical parameters of the power grid, the operation mode of the power grid, the power flow distribution, the load and other factors are taken as the index parameters of the benchmark evaluation of the line loss rate in the transformer district, and the different characteristics of different stations are fully mined, and the static data and dynamic data of the transformer district are extracted respectively, and the abnormal data are corrected by the K nearest neighbor complementarity method. Combined with the characteristic differences of marketing customer portraits, the quantitative feature decomposition of the benchmark value of line loss rate in Transformer district is realized. According to the dynamic identification results of the benchmark value of line loss rate in Transformer district, the benchmark value of line loss rate in Transformer district with high-dimensional data is compressed into a low-dimensional space by using the feature space dimension reduction method, and the quantitative calculation of correlation and difference between samples in each Transformer district is realized. The simulation results show that the calculation complexity of this method is low, and the calculation method has a good ability to adjust the learning rate, ensuring a good local optimal solution in the learning process and high calculation accuracy.

Numerical-analytical method for the Burgers equation

We construct an efficient numerical-analytical method for solving the initial-boundary value problem for the Burgers equation on a segment with a periodic boundary condition. The method includes the reduction to a linear problem based on an explicit-implicit time discretization scheme and an analytical solution of an auxiliary linear problem at each time step using the explicit form of the corresponding Green’s function. The efficiency of the constructed method is due to the fact that the algorithm for solving the auxiliary problem has only linear complexity in terms of the number of spatial discretization nodes used, without involving difference approximations of the derivatives of the desired function. On the basis of the Cole-Hopf substitution, we obtain an explicit periodic solution of the problem on the interval and compare the results of the numerical implementation of the constructed algorithm with this explicit solution. The developed method demonstrated a combination of high computational efficiency and accuracy of the result.

Investigation of the compression-bending capacity evaluation of non-bolted segmental joints

This study aims to thoroughly investigate the compression-bending capacity of non-bolted segmental joints (without bolt connection or connect bolt failure) and provide a basis for its compression-bending safety evaluation. First, a theoretical model regarding the compression-bending capacity of non-bolted segmental joints was proposed and verified by full-scale failure tests. Second, a comparison between bolted and non-bolted segmental joints on compression-bending capacity was conducted. Third, the influence of the segment thickness and length of the core pressure zone on the compression-bending capacity of non-bolted joints were analyzed based on the proposed theoretical model and joint parameters from six segmental joints with different segment thicknesses, and the basic characteristic compression-bending capacity curves of non-bolted joints were obtained. Finally, for the convenience of engineering applications, a simplified method for calculating the compression-bending capacity of non-bolted segmental joints was proposed. The results showed that the relative errors between the theoretical model and full-scale tests under positive and negative bending were 3.4% and 6.1%, respectively, indicating that the theoretical model had high calculation accuracy. The influence of the bolt on the compression-bending capacity of a segmental joint increased with increasing segment thickness and decreasing axial force. With the same eccentricity, the compression-bending capacity of the non-bolted segmental joint generally increased with the segment thickness. The compression-bending capacity decreased with increasing length of the core pressure zone under positive bending, while under negative bending the reverse is true. The characteristic compression-bending capacity curve of the non-bolted joint passed through the origin and had three characteristic points. There was a basic compression-bending capacity curve that was independent of the joint parameters for non-bolted joints with different segment thicknesses. The proposed simplified method for calculating the compression-bending capacity of non-bolted joints is highly accurate and easy to use.

Node's residual descent method for linear elastic boundary value problems

This study proposes a node's residual descent method (NRDM) for the linear elasticity boundary value problem. In this method, the boundary information and deformation states are attached to the nodes, and the partial differential equations (PDEs) are then transformed into a generalized difference form with the first-order generalized finite difference (GFD) scheme. Furthermore, this study thoroughly describes the NRDM, confirms the accuracy and convergence by numerical examples, performs an error analysis, and discusses the influence of computational parameters. Results demonstrate that NRDM has high computational accuracy and superliner convergence rate. The double derivation is implemented to solve the second-order PDE boundary value problem with the first-order GFD scheme, thus simplifying the requirement for star connectivity and reducing the computational complexity. NRDM is a matrix-free iterative method that eliminates the task of assembling large global sparse matrices, thus avoiding the numerous matrix–matrix product operations. Rather than solving large systems of linear equations, many extremely small matrix multiplications are present during iterations, which are extremely friendly to parallel computing and can be adapted to the computation of ultra-multi-degree-of-freedom problems. The computational accuracy and convergence rate are considerably improved in NRDM compared to the classical GFD method.