Search

In this article, for the first time, we propose channel characteristic-based deep neural network (DNN) models for accurate eye-height (EH) and eye-width (EW) estimation of high bandwidth memory (HBM) silicon interposer channels. The proposed models preprocess the design parameters that are highly relevant to the characteristics of the HBM channels. By taking account of the contribution of each design parameter to the eye diagram, the proposed models can accurately estimate the EH and EW of the channels even with a limited number of datasets. For verification, the proposed DNN models were applied to the microstrip and stripline channels of an HBM silicon interposer. Only redistributed layer (RDL) was used to clearly see the effect of reflecting the channel characteristics of the proposed method. We compared the proposed DNN models with various regression methods and a conventional fully connected multilayer DNN model. As a result, the proposed DNN models reduced the EH and EW error rates by 22.7 and 43.9% compared to the other regression methods. In addition, the proposed DNN models not only reduced the error rates by 22.0–28.4% but also reduced the computing cost by 8.0–9.4%, compared to the conventional DNN model. Moreover, we compared the proposed models with various DNN models having other preprocessing structures. By showing 26.7 and 28.8% lower EH and EW error rates than the other DNN models, we validated that the proposed models properly consider the most dominant design factors in preprocessing.

In this article, we propose a near field analytical model of an electromagnetic interference reduction method using an n th harmonic frequency shielding coil with efficiency enhancement in a loosely coupled automotive wireless power transfer system. In general, in the loosely coupled automotive wireless power transfer system, the power transfer efficiency is relatively lower than a tightly coupled wireless power transfer system due to the low coupling. In addition, the electromagnetic interference at the n th harmonic frequency from the power source can interfere with nearby electronic devices. To reduce electromagnetic interference level and enhance efficiency, we proposed a shielding coil between a transmitter and a receiver coil. By using a near field analytical model of the magnetic field intensity and efficiency equations, we analyze based on the equations and obtain a proper value of the resonant frequency of the shielding coil. Therefore, we can achieve low levels of electromagnetic interference at the specific n th harmonic frequency and higher value of the power transfer efficiency in the loosely coupled automotive wireless power transfer system.

An air compressor has been adapted to a high-speed train to reduce the aerodynamic problems when passing through a tunnel. To evaluate the feasibility of the air compressor to attenuate the aerodynamic drag, pressure transient and the micro-pressure wave, the flow around the high-speed train has been simulated using the unsteady axisymmetric compressible Navier–Stokes equation. The simulation results by the dedicated CFD program for the train-tunnel aerodynamics named C-STA™ showed good agreements with the field test result of Korea Train eXpress train running through the Bok-An tunnel. By adapting a compressor with 20% of the train cross-sectional area, the simulation results showed 9% reduced average aerodynamic drag, 54% reduced peak pressure rise in tunnel, and 20% reduced micro-pressure wave, which prove that adapting a compressor can be a good measure to attenuate the tunnel aerodynamic problems for high-speed train.

Usually, modeling takes a long time because it depends on the engineer's experience and is done through repetitive tuning. In this paper, we propose a deep neural network (DNN)-based lumped circuit modeling method using an impedance curve. The proposed method provides a fast and accurate electrical circuit model of inductance (L), capacitance (C), and conductance (G) using a DNN. Since the LCG parameters are predicted by the impedance curve, it is flexible for various applications. For accurately predicting lumped circuit parameters, the DNN model is designed and trained through various case studies. As a result, the proposed method predicts 100% accuracy in inductance and conductance, and 92% accuracy in capacitance. In other words, the proposed method successfully models the electrical characteristics of various applications.

In this paper, we first propose a policy gradient reinforcement learning (RL)-based optimal decoupling capacitor (decap) design method for 2.5-D/3-D integrated circuits (ICs) using a transformer network. The proposed method can provide an optimal decap design that meets target impedance. Unlike previous value-based RL methods with simple value approximators such as multi-layer perceptron (MLP) and convolutional neural network (CNN), the proposed method directly parameterizes policy using an attention-based transformer network model. The model is trained through the policy gradient algorithm so that it can achieve larger action space, i.e. search space. For verification, we applied the proposed method to a test hierarchical power distribution network (PDN). We compared convergence results depending on the action space with the previous value-based RL method. As a result, it is validated that the proposed method can cover ×4 times larger action space than that of the previous work.

In this paper, we proposed a deep neural network (DNN)-based estimation of efficiency enhancement by an intermediate (Int) coil in automotive wireless power transfer (WPT) system. The Int coil can enhance the efficiency in the WPT system with the proper resonant frequency of the Int coil. The previous study has explained the resonant frequency of the Int coil should be higher than the operating frequency. According to the resonant frequency of the Int coil, we can achieve the amount of efficiency enhancement. Therefore, the design of the Int coil is essential for optimize the efficiency enhancement of the automotive WPT system. However, it is impossible to achieve the optimize results of efficiency enhancement by simulations. The proposed DNN-based estimation method can predict the amount of the efficiency enhancement in real cases consisted of ferrites and shielding structures.

In this paper, we proposed a deep neural network (DNN)-based estimation of EMI reduction by an intermediate (Int) coil in automotive wireless power transfer (WPT) system. The Int coil can reduce electromagnetic interference (EMI) level in the WPT system with the proper resonant frequency of the Int coil. The previous study has explained the resonant frequency of the Int coil should be higher than the operating frequency. According to the resonant frequency of the Int coil, we can achieve the amount of EMI reduction. Therefore, the design of the Int coil is essential for optimization of the EMI reduction of the automotive WPT system. However, it is impossible to get the optimized results of EMI reduction by simulations. The proposed DNN-based estimation method can predict the amount of reduced EMI level in real cases consisted of ferrites and shielding structures.

In this paper, we propose an induction heating system analysis method under operating conditions. In the induction heating system, the current of the coil generates an eddy current on the load surface, and heat is generated by the eddy current. At this time, a high level of current is applied depending on the operating conditions of the induction heating system. When the equivalent resistance and inductance of the coil and load are measured and applied to the time domain simulation in the stop condition, the output voltage and current show different results from the measured results. Therefore, we propose a simulation method that calculates the equivalent resistance and inductance in the operating conditions of the induction heating system, and verify the proposed method by measurements. When the proposed method is applied to the simulation, the errors of the output voltage and output current are reduced up to 90% compared to the conventional method when the time domain simulation results are compared with the measurements.

The development of the electric vehicle industry has changed the concept of automobiles as electronic devices. Automotive components are becoming more and more electronic. To ensure the reliability and safety of electronic devices with electric shock and electromagnetic compatibility, each part has protective components such as electrostatic discharge devices and fuses. In particular, as the voltage and current of high-performance vehicle engines increase, high-voltage fuses for protecting drivers and passengers from explosion are becoming more important. In addition, the development of electrical fuses is time and cost consuming due to complex multiphysics, including electric arcs and large-scale experimental environments. In this paper, we propose a model to estimate the voltage, current and internal pressure of a high voltage fuse. Using the concept of a specific action, the resistance of the fuse to induced energy is modeled. The internal pressure is modeled based on the estimated electrical power profile. The proposed model is validated with time-domain measurements and Ansys CFX simulation results. The proposed model can significantly save simulation time and it has the expandability to estimate multiple physics parameters in fuses.

In this paper, we first propose a fast and accurate deep neural network (DNN) model extension method for signal integrity (SI) applications. Reusing pre-trained weights of DNN model, the model can be extended when new training data are given. Instead of updating whole weights of DNN in traditional machine learning (ML) approaches, fine-tuning of a part of weights can accelerate training. For verification, we applied the proposed method to regression model of peak time domain reflectometry (TDR) impedance of through hole via (THV) and classification model of through silicon via (TSV) void defects. Training time of the proposed method were 0.3 s and 2.3 s respectively, which are 99 % and 82.3 % reduction compared to the traditional approach. Moreover, test accuracy of the proposed method achieved 99.2 % and 100 %, respectively.

In this paper, we propose a noise coupling analysis method for electronic safety and arming device (ESAD) based on a 3D-electromagnetic simulation. To analyze dominant noise coupling path in the ESAD, we propose two types of simulation methods. Theses method are based on electromagnetic field detection probes. First simulation method is to ground the target victim signal trace to the ground plane. The other method is to make the target victim signal path open circuit. Based on the proposed simulation methods, we can clarify dominant noise coupling path in the PCB of ESAD. For verification, we applied proposed simulation method on various case that can affect capacitive and inductive noise coupling in the PCB. When the dielectric constant and the conductivity of the PCB were changed, the conductivity only affected noise coupling at frequency range where inductive coupling was dominant, and the change of dielectric constant of PCB was affected at frequency range where capacitive coupling was dominant.

As technology advanced, the demand for data bandwidth has been increasing. To meet this demand, the data rate of the channel has been increased, which causes a lot of signal integrity problems. Optimization of the channel is important to solve these problems. Channels are usually optimized with the empirical knowledge of signal integrity designers. However, it is not accurate and requires numerous iterations. On the other hand, a Bayesian optimization method can quickly find optimized parameter values without relying on empirical knowledge. Therefore, this paper proposes a method for optimizing high-speed channel using Bayesian optimization. The proposed method optimizes the frequency response result such as insertion loss, and find the optimal physical dimension parameters. Finally, the optimized results of the proposed method were verified by comparing all the simulation results in the range of the channel.

A modified method has been suggested to improve the computational efficiency of a multiresolution analysis for implicit temporal integrations. The fundamental idea behind multiresolution analysis is that high cost computations such as flux evaluation are performed only at the cells where the salient flow features exist, thereby reducing the computational cost. The multiresolution analysis, however, suffers from numerical inefficiency in implicit time integration because the efficiency of the multiresolution analysis method arises mainly from a reduction in the expensive flux evaluation for spatial discretization. This paper demonstrates that the implicit temporal integration of multiresolution analysis can be substantially improved by reducing the implicit operator matrix to the size of the additional residual dataset. To suppress the numerical error from the reduced implicit operation, the operator matrix is derived based on the additional residual dataset constructed by the decomposition and thresholdin...

An electronic safety and arming (ESAD) system has a capacitor discharging circuit (CDC) to detonate a fuse. The CDC uses over thousands voltage to ensure the reliability of detonation. When a capacitor of CDC is discharged, a very high and broadband current discharged from the CDC causes noise coupling to noise-sensitive peripheral circuits. Therefore, in order to guarantee operational reliability of the ESAD, it is necessary to analyze noise coupling from the high current generated by the CDC. In this paper, we analyze noise coupling in the ESAD system. Noise coupling paths between CDC and trigger circuit are analyzed with full wave 3D-electromagnetic (EM) simulation. Also, simulation in the time domain is conducted to verify the voltage noise induced on the trigger circuit. We propose two ground designs to suppress noise coupling from the CDC. The effectiveness of the different ground designs is compared with time domain simulation. As a result, the proposed ground designs suppress the voltage noise induced on the trigger circuit up to 56 %.

As technology develops, the amount of data used increases exponentially and the operating speed of electronic devices increases accordingly. As the operating speed increases, the transmission speed of data exchanged through the channel also increases. To get an eye-diagram, a time-domain simulation is required, which consumes a lot of time and computer power. However, a channel simulation in frequency-domain is fast and it consumes less computer power. Therefore, using the frequency-domain simulation data is more efficient than the time-domain simulation. This paper proposes an eye-width and eye-height estimation method using an artificial neural network (ANN). The input of the ANN is insertion loss and the outputs of the ANN are eye-width and eye-height. Finally, the performance of the proposed method is verified by transient eye-diagram simulations with arbitrarily-selected channel parameters.

SUMMARY A multi-resolution analysis (MRA) is proposed for efficient flow computation with preserving the high-order numerical accuracy of a conventional solver. In the MRA process, the smoothness of a flow pattern is assessed by the difference between original flow property values, and the values approximated by high-order interpolating polynomial in decomposition. Insignificant data in smooth region are discarded, and flux computation is performed only where crucial features of a solution exist. The reduction of expensive flow computation improves the overall computational efficiency. In order to maintain the high-order accuracy, modified thresholding procedure restricts the additional error introduced by the thresholding below the order of accuracy of a conventional solver. The practical applicability of the MRA method is validated in various continuous and discontinuous flow problems. The MRA stably computes the Euler equations for continuous and discontinuous flow problems and maintains the accuracy of a conventional solver. Overall, it substantially enhances the computational efficiency of the conventional third-order accurate solver. Copyright © 2014 John Wiley & Sons, Ltd.

As the preceding research for the design of gangway in the next generation high speed train, the aero-acoustic noise at the gangway is calculated. For this purpose, the shape of gangway with mud flaps is assumed as the two-dimensional cavity. Then, 5 gap sizes between mud flaps of gangway are selected and parametric study is performed according to the gap sizes. From this study, the aerodynamic features such as vortex shedding, pressure, etc. are computed. Also, the aero-acoustic properties of tonal noise and overall noise are analyzed at the 3 locations of microphone and the relation between the gap size of mud flap and the noise level is assessed. Through this study, it is shown that the noise characteristics of base and specific models are better than those of other models.

Reliability-based design optimization (RBDO) of the NASA stage 37 axial compressor is performed using an uncertainty model for stall margin in order to guarantee stable operation of the compressor. The main characteristics of RBDO for the axial compressor are summarized as follows: First, the values of mass flow rate and pressure ratio in stall margin calculation are defined as statistical models with normal distribution for consideration of the uncertainty in stall margin. Second, Monte Carlo Simulation is used in the RBDO process to calculate failure probability of stall margin accurately. Third, an approximation model that is constructed by an artificial neural network is adopted to reduce the time cost of RBDO. The present method is applied to the NASA stage 37 compressor to improve the reliability of stall margin with both maximized efficiency and minimized weight. The RBDO result is compared with the deterministic optimization (DO) result which does not include an uncertainty model. In the DO case, stall margin is slightly higher than the reference value of the required constraint, but the probability of stall is 43%. This is unacceptable risk for an aircraft engine, which requires absolutely stable operation in flight. However, stall margin obtained in RBDO is 2.7% higher than the reference value, and the probability of success increases to 95% with the improved efficiency and weight. Therefore, RBDO of the axial compressor for aircraft engine can be a reliable design optimization method through consideration of unexpected disturbance of the flow conditions.

The demands for small, high performance and high loaded aircraft compressor are increased in the world. But the design requirements become increasingly complex to design these high technical engines, the requirement of the design optimization become increased. The optimal design result of several disciplines show different tendencies and nonlinear characteristics of the compressor design, the multidisciplinary design optimization method must be considered in compressor design. Therefore, the artificial Neural Net method is adapted to make the approximation model of 3-stage axial compressor design optimization for considering the nonlinear characteristic. At last, the optimal result of this study is compared to that of previous study.

In this study, an experiment was performed to obtain the optimum design of a passive micromixer for effective mixing by using a microsized device and rectangular obstacles; a low Reynolds number was maintained in the microchannel. The experiment was carried out by varying the number, size, and location of the rectangular obstacles. Further, the Y-channel's shape was optimized for maximizing the mixture ratio, which has limit qualification that an allowed value of pressure drop. The increase in the efficiency of mixing was observed to be greater than that in the case of circular obstacles by approximately 2.5%.

An enhanced Multi-dimensional Limiting Process (e-MLP) is developed for the accurate and efficient computation of multi-dimensional flows based on the Multi-dimensional Limiting Process (MLP). The new limiting process includes a distinguishing step and an enhanced multi-dimensional limiting process. First, the distinguishing step, which is independent of high order interpolation and flux evaluation, is newly introduced. It performs a multi-dimensional search of a discontinuity. The entire computational domain is then divided into continuous, linear discontinuous and nonlinear discontinuous regions. Second, limiting functions are appropriately switched according to the type of each region; in a continuous region, there are no limiting processes and only higher order accurate interpolation is performed. In linear discontinuous and nonlinear discontinuous regions, TVD criterion and MLP limiter are respectively used to remove oscillation. Hence, e-MLP has a number of advantages, as it incorporates useful features of MLP limiter such as multi-dimensional monotonicity and straightforward extensionality to higher order interpolation. It is applicable to local extrema and prevents excessive damping in a linear discontinuous region through application of appropriate limiting criteria. It is efficient because a limiting function is applied only to a discontinuous region. In addition, it is robust against shock instability due to the strict distinction of the computational domain and the use of regional information in a flux scheme as well as a high order interpolation scheme. This new limiting process was applied to numerous test cases including one-dimensional shock/sine wave interaction problem, oblique stationary contact discontinuity, isentropic vortex flow, high speed flow in a blunt body, planar shock/density bubble interaction, shock wave/vortex interaction and, particularly, magnetohydrodynamic (MHD) cloud-shock interaction problems. Through these tests, it was verified that e-MLP substantially enhances the accuracy and efficiency with both continuous and discontinuous multi-dimensional flows.

This study seeks to validate the accuracy and the efficiency of the aerodynamic reduced order model (ROM). In doing this, snapshot data are generated from the full system analysis of a fighter wing problem. From an eigensystem analysis of these snapshots, the basis vector reproducing the behavior of the full system is obtained. The span length, sweep angle, dihedral angle, and spar and rib thickness representing the wing configuration are determined as the input variables. The constructed ROM is applied to the fighter wing problem while varying the input conditions for validation. Subsequently, a comparison of the reduced system with the full system confirmed that the aerodynamic performance is within 4% error and that the L2 norms are 10−6 order of the entire flow field. Therefore, the ROM is able to capture the variation of the aerodynamic performance with respect to the input variables. Though there are structural input variables which influence the aerodynamic performance indirectly, the ROM can reproduce the flow field of the full system. Additionally, even if the ROM incurs a high computational cost to generate snapshots, it can represent the behavior of the full system efficiently once the reduced order model is constructed.

Sanghyun Chae Department of Aerospace Engineering, Pusan National University, Busan, 609-735, Korea Hyungmin Kang Department of Mechanical Engineering, Dongyang Mirae University, Seoul, 152-714, Korea Sejong Oh Department of Aerospace Engineering, Pusan National University, Busan, 609-735, Korea and Kwanjung Yee Department of Mechanical & Aerospace Engineering, Seoul National University, Seoul, 151-742, Korea

Supercompact wavelets are extended to hybrid-type applications for practical fluid dynamics data management. Beam and Warming's supercompact multiwavelet method is regarded as one proper tool for representing fluid simulation data because of its compact support. This wavelet compactness significantly reduces storage requirements as well as data processing time, and can prevent unnecessary interaction with closely located data across singularities such as shocks and vortices. For effective domain subdivision, a hybrid method that combines 3-D, 2-D, and 1-D versions of supercompact multiwavelets is developed. The hybrid 3-D type of supercompact multiwavelet shows high data compression ratio in practical fluid simulation data because the mother wavelets are highly effective and encompass subdivision methods. With the present method, efficient data management for datasets of huge size is enabled without losing important features. Several numerical tests substantiate large data compression for flow fi...

A hybrid method with supercompact multiwavelets is suggested as an efficient and practical method to compress CFD dataset. Supercompact multiwavelets provide various advantages such as compact support and orthogonality in CFD data compression. The compactness is a crucial condition for approximated representation of CFD data to avoid unnecessary interaction bet­ween remotely spaced data across various singularities such as shock and vortices. But the supercompact multi wavelet method has to fit the CFD grid size to a product of integer and power of two. m×2^n. To resolve this problem, the hybrid method with combination of 3.2 and 1 dimensional version of wavelets is studied. With the hybrid method, any arbitrary size can be handled without any shrinkage or expansion of the original problem. The presented method allows high data compression ratio for fluid simulation data. Several numerical tests substantiate large data compression ratios for now field simulation successfully.

An experimental study has been performed in an attempt to seek some possibilities of obtaining the solutions of heat transfer problems related to rotor-stator disk cavity systems of gas turbine engines by analogy with corresponding problems in mass transfer which arises from using naphthalene sublimation technique. Measurements are made to examine the effects of rotational Reynolds number, the flow rate and the gap ratio on the radial pressure distributions. Pressure inversion phenomenon has been verified to exist for the case of shrouded disks with radial clearance, which perform better in terms of heat transfer, too, than those with axial clearance. The stator flow reattachment points are determined directly by the dual sensor pressure probes and compared well with those estimated from the distributions of static pressure and mass transfer coefficients.

The adaptive wavelet method is studied for the enhancement of computational efficiency of three-dimensional flows. Three-dimensional wavelet decomposition process is introduced based on the previous two-dimensional method. The order of numerical accuracy of an original solver is preserved by applying a proper thresholding technique. The proposed algorithm is applied to the computation of flow field around ONERA-M6 wing in transonic regime. Through the application, it is confirmed that the three-dimensional adaptive wavelet method can reduce the computational time with conserving the numerical accuracy of an original solver.

An adaptive wavelet method is applied in order to enhance the computational efficiency of enhanced Multi-dimensional Limiting Process (e-MLP) without deterioration of the numerical accuracy of original Computational Fluid Dynamics (CFD) scheme. For this purpose, higher order of adaptive wavelet method is constructed including higher order of wavelet decomposition and modified thresholding. Besides, the locations of crucial features such as shock, vortex core, etc. are automatically and accurately searched in the CFD dataset through wavelet transformation. Only on these locations, high order spatial interpolation scheme with e-MLP are performed; in the other locations, interpolation method is utilized to compute residual values, which reduces the computational time of flux evaluation. This high order adaptive wavelet method was applied to unsteady Euler flow computations such as shock-vortex interaction problem. Throughout these processes, it was verified that computational efficiency was enhanced with preservation of numerical accuracy of CFD solver.Copyright © 2011 by KSME

An adaptive wavelet method is proposed to achieve an efficient algorithm with maintaining numerical accuracy of a conventional solver in case of high order accurate Euler computations. Due to wavelet transformation including wavelet decomposition and thresholding, dataset is automatically adapted to local features of a solution. Then, the flux values are calculated only at the adapted dataset, which enables the enhancement of the computational efficiency. For maintaining the higher order accuracy of a conventional solver, threshold value is adjusted to consider the spatial and temporal accuracies. Through this threshold value, the deterioration of the numerical accuracy due to thresholding is obstructed and the wavelet method can present high order accurate solutions. The adaptive wavelet method was successfully applied to high order accurate Euler flow computation. In consequence, computational efficiency was enhanced while maintaining the higher order numerical accuracy of a solution.

In medical imaging, the use of three-dimensional image is increasing for both educational and diagnostic purposes. With the advent of techniques for browsing a three-dimensional object, it became possible to display three-dimensional images on a personal computer via the Internet. This report describes the construction of a three-dimensional virtual model of human lower extremities that was linked with the data acquired from gait analysis in order to perform three-dimensional gait analysis.

In Computational Fluid Dynamics (CFD), a variety of high order schemes have been extensively applied to an integrated design process, however its capacity still fall short of the full scale integrated design. Full scale integrated design requires a large number of grid points in spatial domain, entailing substantial increase of computing time. However, it is a waste of computational resources to use fine grid in the whole domain. A dense grid is needed only in a rapidly changing region such as shock waves, boundary layers, etc. For that reason, several types of adaptive methods have been implemented to improve efficiency of analysis tool. In these days, numerical techniques based on wavelets methods have emerged as one of revolutionary technologies.

This paper introduces a novel method to control the tilting motion of an optical micromirror using EWOD (Electrowetting-on-Dielectric) actuation that can be used in optical devices (e.g., optical scanner, large scale display). Using the vertical motion of a liquid droplet during EWOD movement that gives a height difference due to the contact angle variation at the three-phase contact line by an applied voltage, a dramatically simplified and effective tilting motion of micromirror can be realized. The actuation concept, fabrication, and test of the initial devices are reported. A preliminary time response experiment of our initial device is also included to examine the feasibility of our device as a large display component.

Data compression technique is proposed that co mbines 3D, 2D and 1D versions of supercompact. Computation domain is divided into subsets for each dimensional version of multiresolution appl icatio n. Compared to Harten's interpolatory mu ltiresolution, supercompact wavelets require small number of support points, lea d- ing to quite simple and less expensive numerical i m- plementation. The hybrid multiresolution method a l- lows high actual data compression for fluid simulation data since the method is based on highly effective wavelets and subdivision methods. Several numerical tests substantiate large data co mpression ratios for flow field simul ation.