Fast Prediction Method Research Articles

Prediction method of successive commutation failure for multi-infeed high voltage direct current systems under fault of weak receiving-end grid

The complex interactions between AC/DC systems and different line-commutated converter-based high-voltage direct current (LCC-HVDC) expand the effect of commutation failure (CF). The rapid prediction of CF is crucial in ensuring the stable operation of such systems. The simultaneous CF caused by grid fault has received widespread attention in recent years. However, the coupling between the LCC-HVDC with CF and the adjacent normal LCC-HVDC is not considered. When the state of the main circuit suddenly changes and the control system subsequently starts, the LCC-HVDC with CF inevitably produces an effect on the adjacent LCC-HVDC through the coupling of grid voltage. This study describes a new phenomenon, where the former produces a successive CF in the latter under the fault of weak receiving-end grid. The dynamic reactive power characteristics of inverter station and the induced mechanism of successive CF are analyzed. And the dynamic reactive power model of LCC-HVDC is established. A fast prediction method for successive CF is then proposed by calculating the extinction angle of LCC-HVDC with consideration of the effect of adjacent LCC-HVDC with CF. The proposed method can improve the accuracy of CF prediction in multi-infeed LCC-HVDC systems through theoretical derivation and simulation verification.

Prediction of Sound Scattering from Deep-Sea Targets Based on Equivalence of Directional Point Sources

A fast prediction method is proposed for calculating the sound scattering of targets in the deep-sea acoustic channel by equating the sound scattering field of a complex elastic target to the acoustic field excited by a directional point source. In deep-sea conditions, the effects of the sea surface on the impedance characteristics of the elastic target surface can be ignored. Through the finite element simulation of the acoustic scattering of the target in the free field, the sound scattering field is equated to the radiation field of a directional point source. Subsequently, the point source is placed in the channel, and the acoustic ray method is used to calculate the distribution of the scattering field. On the basis of theoretical modelling, the method of obtaining the directional point source and the influence of the sea surface on the impedance of the scattering field are analysed. Subsequently, the proposed method is compared with the finite element method in terms of computational efficiency. The result shows that the method considers the multiple complex coupling effects between the elastic structure and marine environment. The influence of the boundary is approximately negligible when the distance from the ocean boundary to the elastic structure is equal to the wavelength. The method only performs finite element coupling calculation in the free field; the amount of mesh size is greatly reduced and the calculation efficiency is significantly improved when compared with the finite element calculation in the entire channel, the. The calculation time in the example can be reduced by more than one order of magnitude. This method organically combines the near-field calculation with acoustic ray theory and it can realise the rapid calculation of the large-scale acoustic scattering field in complex marine environments.

A reduced-order model based on finite element method for fast prediction of thermal performance of lattice structures

Lattice structures, emerging with additive manufacturing technology, have broad application prospects in industry, and desirable thermal and mechanical properties of which can be obtained by changing the internal cells and their array mode. However, the computational cost for the design and analysis of a lattice structure is usually extremely high due to the structural complexity. For fast prediction of thermal performance of lattice structures, a reduced-order model was developed, in which the complex lattice structure is simplified into a three-dimensional rod system and the heat conduction, convection, and radiation are modeled in a reduced second-order way based on the finite element principle. The convergence and accuracy was examined, and the model validation was done by comparing with the corresponding full model. The results showed that the calculation error of the reduced-order model decreases as the rod aspect ratio increases and the computational efficiency is improved by more than 1000 times. The reduced-order model was then applied to evaluate the thermal insulation and thermal dissipation performance of seven kinds of lattice cells, and their characteristics of the thermal performance were found and compared. We finally gave a better designed cell, which shows the best thermal insulation and dissipation performance simultaneously.

Context-Based Inter Mode Decision Method for Fast Affine Prediction in Versatile Video Coding

Versatile Video Coding (VVC) is the most recent video coding standard developed by Joint Video Experts Team (JVET) that can achieve a bit-rate reduction of 50% with perceptually similar quality compared to the previous method, namely High Efficiency Video Coding (HEVC). Although VVC can support the significant coding performance, it leads to the tremendous computational complexity of VVC encoder. In particular, VVC has newly adopted an affine motion estimation (AME) method to overcome the limitations of the translational motion model at the expense of higher encoding complexity. In this paper, we proposed a context-based inter mode decision method for fast affine prediction that determines whether the AME is performed or not in the process of rate-distortion (RD) optimization for optimal CU-mode decision. Experimental results showed that the proposed method significantly reduced the encoding complexity of AME up to 33% with unnoticeable coding loss compared to the VVC Test Model (VTM).

Massively Parallel GPU-Accelerated String Method for Fast and Accurate Prediction of Molecular Diffusivity in Nanoporous Materials

The diffusivity of guest molecules in nanoporous materials is instrumental for practical applications ranging from gas separation to catalysis and energy storage. Conventional methods to predict di...

Fast prediction method of aircraft noise with distributed propulsion in the far field

Based on the known methods for calculating the propeller noise, a fast prediction method of the distributed propulsion noise is proposed. The method takes into account only the component of tonal noise from the load. To test the adequacy of the model, the GL-10 model was used, a concept of an aircraft developed by National Aeronautics and Space Administration to test distributed propulsion as well as vertical take-off and landing maneuvers. The comparison showed a good agreement between the results obtained and the data for the GL-10 model.

Improving Performance Estimation for Design Space Exploration for Convolutional Neural Network Accelerators

Contemporary advances in neural networks (NNs) have demonstrated their potential in different applications such as in image classification, object detection or natural language processing. In particular, reconfigurable accelerators have been widely used for the acceleration of NNs due to their reconfigurability and efficiency in specific application instances. To determine the configuration of the accelerator, it is necessary to conduct design space exploration to optimize the performance. However, the process of design space exploration is time consuming because of the slow performance evaluation for different configurations. Therefore, there is a demand for an accurate and fast performance prediction method to speed up design space exploration. This work introduces a novel method for fast and accurate estimation of different metrics that are of importance when performing design space exploration. The method is based on a Gaussian process regression model parametrised by the features of the accelerator and the target NN to be accelerated. We evaluate the proposed method together with other popular machine learning based methods in estimating the latency and energy consumption of our implemented accelerator on two different hardware platforms targeting convolutional neural networks. We demonstrate improvements in estimation accuracy, without the need for significant implementation effort or tuning.

Efficient Graphene Reconfigurable Reflectarray Antenna Electromagnetic Response Prediction Using Deep Learning

Aiming at the time-consuming problem of the full-wave (FW) simulation of the scattering characteristics of the traditional graphene reconfigurable reflectarray antenna, a fast prediction method of electromagnetic (EM) response based on deep learning is proposed. The convolutional neural network (CNN) method in deep learning is effectively used in the research of this paper. This method first discretizes the input vector (patch geometry, chemical potential, frequency, incident angle, etc.) of the graphene reflectarray antenna, and then preprocesses the data into a two-dimensional image suitable for CNN training, and finally uses CNN to train the model instead of extensive FW simulation calculations, the EM response of the reflectarray antenna is calculated. The training results of three algorithms of support vector regression (SVR), radial basis function network (RBFN) and CNN are comprehensively compared. The experimental results show that CNN method has good performance and accuracy in the EM response prediction of the graphene reconfigurable reflectarray antenna, with an accuracy of over 99%, and can also save at least 99% of time.

A qualitative characteristic scheme and a fast distance prediction method of multi-probe azimuthal gamma-ray logging in geosteering

Azimuthal gamma-ray logging instruments play an important role in the geological steering technology (geosteering) of directional drilling engineering. While their logging response characteristics in various geology bodies are ambiguous, and a fast prediction methods of the distances between logging instruments and formation boudnaries, which are suitable for real time geosteering in drilling, are rare. In order to settle these problems, a Monte Carlo model is established based on the two-layer model and the multi-probe azimuthal gamma-ray logging (MPAGR) instrument structure. Monte Carlo N-Particle Transport (MCNP) Code is employed to simulate the logging responses of MPAGR when drilling into various geology structure bodies (layerd formations, fault, pinchouts and lens), which commonly drilled by honrizontal wells, and then a logging response characteristic scheme in eight spatial relationships between horizontal wells and formations is analyzed and summerized for guiding geosteering qualitatively. Based on these numerical simulations, a fast prediction method, a series of non-linear functions which can describe the relationships between logging responses and boundary distances, is proposed to guide realtime geosteering quantitatively. Finally, the qualitative scheme and the quantitative prediction method are applied in two horizontal wells in the Songliao Basin, China. The application results and effects demonstrate the availability of the scheme and the reliability of the fast prediction method. The scheme and the prediction method can provide reference and alternative solutions for real-time geosteering when drilling horizontal wells, especially in the formations where the geosteering function of resistivity logging is limited, e.g., shale gas/oil, coalbed methane, thin layer and low resistivity contrast reservoirs.

스마트 기술을 활용한 승강기 위험 상황 자동 예측판단 기법 개발

In Korea, the first elevator was introduced in 1910. Thereafter, Korea has been ranked 3rd in installation and 7th in maintenance worldwide. During the last five years, an average of 35 000 elevators per year have been installed, and the total number of installed elevator increases every year. In Korea, 92.1 % of the elevator malfunctions are to be stuck in an elevator. When passengers are stuck in an elevator, elevator maintenance personnel or firefighters usually rescue them; however, the number of firefighter rescues has been increasing compared to the number of new elevators installed. Along with the increase in being stuck in an elevator and the increase in the number of firefighter rescues, anxiety of elevator passengers and social costs are also increasing. Hence, there is a need to find a method to reduce these incidences. Therefore, it is essential to develop an intelligent rescue system using video, video, vibration, noise and elevator’s malfunction signal for various hazard pattern recognition. In this study, we propose a new technology that can prevent incidents such as being stuck in an elevator, violence or collapse, and minimize the gap in field management through a fast and accurate hazard prediction and response method based on machine learning.

Machine Learning-Based Radio Coverage Prediction in Urban Environments

AIM: Having a reliable prediction model of radio signal strength is an essential tool for planning and designing a radio network. Given a geographic region, and associated power estimates linked to the transmitter placements, our objective is to develop machine learning models to predict the strength of the radio signals. BACKGROUND: The propagation model is often used to determine the optimal location of radio transmitters in order to optimize the power coverage in a geographic area of interest. However, it is often a costly operation to obtain the exact power measurements over a region for a given set of transmitter locations. Therefore, fast prediction methods are needed to estimate the power values given limited data. METHODOLOGY: We consider a dataset consisting of simulated power at each point in an environment for a given set of transmitter locations. We experiment with various machine learning models, namely, generalized linear models (GLMs), neural networks (NNs), and k-nearest neighbor (KNN), to estimate the power values for a given transmitter placement. We investigate various feature engineering approaches to enhance the predictive performance of the machine learning models. RESULTS: We observe that employed feature engineering methods such as polynomial degrees and transmitter to cluster distances significantly improve the prediction accuracy. In particular, GLM model performance notably improves thanks to these extracted features, where mean absolute error (MAE) is reduced around 77% from 11.37 dB to 2.55 dB. We note that KNN with k = 2 and DNN models perform better than NN and GLM. KNN has the best performance with an average MAE of 0.65dB and also substantially faster to train than NN/DNN models. In addition, our analysis shows that, to train a well-performing machine learning model, it is sufficient to use a dataset consisting of measurements at a fraction of the potential transmitter locations in a given region. CONCLUSIONS: Machine learning methods are highly effective for the coverage prediction task. Using carefully engineered features, simple models such as GLMs and KNNs can be as effective as more complex ones, especially for small datasets.

AliFreeFoldMulti: alignment-free method to predict secondary structures of multiple RNA homologs.

Predicting RNA structure is crucial for understanding RNA’s mechanism of action. Comparative approaches for the prediction of RNA structures can be classified into four main strategies. The three first—align-and-fold, align-then-fold and fold-then-align—exploit multiple sequence alignments to improve the accuracy of conserved RNA-structure prediction. Align-and-fold methods perform generally better, but are also typically slower than the other alignment-based methods. The fourth strategy—alignment-free—consists in predicting the conserved RNA structure without relying on sequence alignment. This strategy has the advantage of being the faster, while predicting accurate structures through the use of latent representations of the candidate structures for each sequence. This paper presents aliFreeFoldMulti, an extension of the aliFreeFold algorithm. This algorithm predicts a representative secondary structure of multiple RNA homologs by using a vector representation of their suboptimal structures. aliFreeFoldMulti improves on aliFreeFold by additionally computing the conserved structure for each sequence. aliFreeFoldMulti is assessed by comparing its prediction performance and time efficiency with a set of leading RNA-structure prediction methods. aliFreeFoldMulti has the lowest computing times and the highest maximum accuracy scores. It achieves comparable average structure prediction accuracy as other methods, except TurboFoldII which is the best in terms of average accuracy but with the highest computing times. We present aliFreeFoldMulti as an illustration of the potential of alignment-free approaches to provide fast and accurate RNA-structure prediction methods.

Serum proteins may facilitate the identification of Kawasaki disease and promote in vitro neutrophil infiltration

Kawasaki disease (KD) usually affects the children younger than 5 years of age and subsequently causes coronary artery lesions (CALs) without timely identification and treatment. Developing a robust and fast prediction method may facilitate the timely diagnosis of KD, significantly reducing the risk of CALs in KD patients. The levels of inflammatory serum proteins dramatically vary during the onsets of many immune diseases, including in KD. However, our understanding of their pathogenic roles in KD is behind satisfaction. The purpose of this study was to evaluate candidate diagnostic serum proteins and the potential mechanism in KD using iTRAQ gel-free proteomics. We enrolled subjects and conducted iTRAQ gel-free proteomics to globally screen serum proteins followed by specific validation with ELISA. Further in vitro leukocyte trans-endothelial model was also applied to investigate the pathogenesis roles of inflammatory serum proteins. We identified six KD protein biomarkers, including Protein S100-A8 (S100A8), Protein S100-A9 (S100A9), Protein S100-A12 (S100A12), Peroxiredoxin-2 (PRDX2), Neutrophil defensin 1 (DEFA1) and Alpha-1-acid glycoprotein 1 (ORM1). They enabled us to develop a high-performance KD prediction model with an auROC value of 0.94, facilitating the timely identification of KD. Further assays concluded that recombinant S100A12 protein treatment activated neutrophil surface adhesion molecules responsible for adhesion to endothelial cells. Therefore, S100A12 promoted both freshly clinically isolated neutrophils and neutrophil-like cells to infiltrate through the endothelial layer in vitro. Finally, the antibody against S100A12 may attenuate the infiltration promoted by S100A12. Our result demonstrated that evaluating S100A8, S100A9, S100A12, PRDX2, DEFA1 and ORM1 levels may be a good diagnostic tool of KD. Further in vitro study implied that S100A12 could be a potential therapeutic target for KD.

Fast prediction method of failure modes for steel box structures under internal blast loading

Steel box structures widely existed in ships, buildings and some steel girder bridges. In our previous work, the failure modes of steel box structures under internal blast loading were observed, summarized and validated through both experimental and numerical studies. On the basis of the previous research, a fast prediction method for the failure modes was proposed in the present study. Firstly, dimensional analyses were carried out for the analysis of failure mode, the main influencing factors were determined and a prediction model was constructed. Then, a large number of numerical simulations considered different kinds of steel, structure dimension, plate thickness and different explosive masses were conducted. The failure modes of all the simulation cases were obtained. After that, failure modes maps on one dimensionless coordinate axis and two-dimensional dimensionless number coordinate system were plotted based on the large numbers of failure modes data, respectively. On which regular distributions of different failure modes were observed and empirical equations were proposed for the fast prediction of failure mode of steel box structure under internal blast loading. In addition, the proposed empirical equation and failure mode map for fast failure mode prediction were verified by experimental results.

Fast prediction for multi-parameters (concentration, temperature and humidity) of indoor environment towards the online control of HVAC system

Heating, ventilation and air conditioning (HVAC) systems are the most energy-consuming building implements for the improvement of indoor environmental quality (IEQ). We have developed the optimal control strategies for HVAC system to respectively achieve the optimal selections of ventilation rate and supplied air temperature with consideration of energy conservation, through the fast prediction methods by using low-dimensional linear ventilation model (LLVM) based artificial neural network (ANN) and low-dimensional linear temperature model (LLTM) based contribution ratio of indoor climate (CRI(T)). To be continued for integrated control of multi-parameters, we further developed the fast prediction model for indoor humidity by using low-dimensional linear humidity model (LLHM) and contribution ratio of indoor humidity (CRI(H)), and thermal sensation index (TS) for assessment. CFD was used to construct the prediction database for CO2, temperature and humidity. Low-dimensional linear models (LLM), including LLVM, LLTM and LLHM, were adopted to expand database for the sake of data storage reduction. Then, coupling with ANN, CRI(T) and CRI(H), the distributions of indoor CO2 concentration, temperature, and humidity were rapidly predicted on the basis of LLVM-based ANN, LLTM-based CRI(T) and LLHM-based CRI(H), respectively. Finally, according to the self-defined indices (i.e., EV, ET, EH), the optimal balancing between IEQ (indicated by CO2 concentration, PMV and TS) and energy consumption (indicated by ventilation rate, supplied air temperature and humidity) were synthetically evaluated. The total HVAC energy consumption could be reduced by 35% on the strength of current control strategies. This work can further contribute to development of the intelligent online control for HVAC systems.

Predicting Inorganic Photovoltaic Materials with Efficiencies >26% via Structure-Relevant Machine Learning and Density Functional Calculations

Summary Discovering new inorganic photovoltaic materials becomes an efficient way for developing a new generation of solar cells with high efficiency and environmental stability. Using machine learning (ML) and density functional theory calculations, we report four promising inorganic photovoltaic materials—Ba4Te12Ge4, Ba8P8Ge4, Sr8P8Sn4, and Y4Te4Se2—demonstrating notable theoretical photovoltaic performance for use in solar cells. The symmetry-allowed optical transition probability, the large amount of density of states near the conduction band minimum (CBM) and the valence band maximum (VBM), and the strong p-p transition across the band edge contribute to the large optical absorption coefficient, leading to the outstanding theoretical power conversion efficiency (PCE). The separation of the VBM and CBM wave function distributions contribute to the fast separation of the photogenerated electrons and holes and the enhanced carrier lifetimes. Our ML model is an efficient method for fast and atomic-level accuracy prediction of photovoltaic materials with different crystal structures.

Interplay Between Switching and Retention in HfO2-Based Ferroelectric FETs

Long data retention is a critical requirement for many of the potential applications of HfO2-based ferroelectric field-effect transistors (FeFETs). However, methods for its rapid assessment are still missing. In this article, we report a detailed investigation of the retention and switching properties of FeFETs fabricated in the 28-nm high- ${k}$ metal gate technology. We identify a clear correlation between the two properties and propose a method for fast prediction of the device retention behavior, which can be easily adopted to judge different fabrication processes. Finally, we extend the validity of the method to a direct assessment of disturb-free operating conditions, which may be particularly valuable for FeFET array operation.

A cavitation performance prediction method for pumps PART1-Proposal and feasibility

Pumps are essential machinery in the various industries. With the development of high-speed and large-scale pumps, especially high energy density, high requirements have been imposed on the vibration and noise performance of pumps, and cavitation is an important source of vibration and noise excitation in pumps, so it is necessary to improve pumps cavitation performance. The modern pump optimization design method mainly adopts parameterization and artificial intelligence coupling optimization, which requires direct correlation between geometric parameters and pump performance. The existing cavitation performance calculation method is difficult to be integrated into multi-objective automatic coupling optimization. Therefore, a fast prediction method for pump cavitation performance is urgently needed. This paper proposes a novel cavitation prediction method based on impeller pressure isosurface at single-phase media. When the cavitation occurs, the area of pressure isosurface Siso increases linearly with the NPSHa decrease. This demonstrates that with the development of cavitation, the variation law of the head with the NPSHa and the variation law of the head with the area of pressure isosurface are consistent. Therefore, the area of pressure isosurface Siso can be used to predict cavitation performance. For a certain impeller blade, since the area ratio Rs is proportional to the area of pressure isosurface Siso, the cavitation performance can be predicted by the Rs. In this paper, a new cavitation performance prediction method is proposed, and the feasibility of this method is demonstrated in combination with experiments, which will greatly accelerate the pump hydraulic optimization design.

A prediction method based on the voxel model and the finite cell method for cutting force-induced deformation in the five-axis milling process

When milling a thin-walled workpiece, knowledge of machining error due to the workpiece deformation induced by cutting force is crucial to ensuring the quality of the final machined part. In the current research of deformation prediction, the finite element method (FEM) is the dominant tool for calculating the deformation induced by the cutting force. However, FEM needs tremendous manual effort to generate a mesh of the in-process workpiece (IPW) and cannot be directly applied to discrete structured models such as the voxel or dexel models that are widely used for its advantage on parallel computation. In addition, a very fine mesh is needed for FEM to preserve the geometry of IPW, but it will lead to a huge global stiffness matrix. Therefore, it calls for a new method of fast deformation prediction in the milling of thin-walled workpieces. In this paper, aiming at the fast prediction of the deformation error of five-axis milling directly and efficiently on the voxel model, we present a prediction algorithm based on a combination of the finite cell method (FCM) and the voxel model. Specifically, in this study, a pipeline of FCM combined with the voxel model is specially designed for the efficient analysis of IPW deformation along a continuous tool path in the five-axis milling of thin-walled workpieces. By introducing a technique of partially updating the global stiffness matrix and the cells’ stiffness matrices, the proposed algorithm can tremendously reduce the FCM calculation time per sample point on the tool path, by as much as 19 times in our test examples. Both computer simulation and physical cutting experiments performed by us have convincingly verified the accuracy of the proposed deformation prediction algorithm. To our best knowledge, this is the first work of introducing FCM on the voxel model to the prediction of deformation induced by cutting force in milling problems, which enables customized tool path design and optimization in the five-axis milling of free-form surfaces.

Developing partial least square (PLS) internal parameters of apple (Malus sylvestris L.) cv. Manalagi by means of UV/Vis spectroscopy

In order to determine the ripening stages of apple (Malus sylvestris L), local farmer in Malang still utilized a traditional method by examining its size and appearance. However, this method needs worker’s extensive experience causing a non-standard ripeness level that may lead to lower the quality of crop. Thus, to strive this problem, fast and quantitative prediction method need to be developed. UV/Vis spectroscopy has shown its capability to provide a robust prediction of several internal attributes such as, Soluble Solid Content (SSC) and moisture content. The aim of this research is to develop a Partial Least Square (PLS) regression to predict internal parameter contained in apple. Fifty sample of apples were taken from local plantation in Bumiaji district. There were 3 stages to complete the research: (1) spectral data acquisition ranging from wavelength 200nm-1100nm; (2) Psycochemical measurement (moisture content, SSC and firmness); (3) Performing PLS regression based on spectral data and internal parameters. The result showed that PLS model of firmness could provide promising result where R2 calibration and validation were 0.658 and 0.635 respectively. On the other hand, the model could not predict both moisture content and SSC resulting 0.57 and 0.308 for R2 calibration. Further enhancement needs to be addressed to the firmness model for improving its capability of prediction.