scispace - formally typeset
Search or ask a question
Author

Ling Zhang

Other affiliations: Zhejiang University
Bio: Ling Zhang is an academic researcher from Missouri University of Science and Technology. The author has contributed to research in topics: Electromagnetic interference & Electromagnetic shielding. The author has an hindex of 6, co-authored 22 publications receiving 67 citations. Previous affiliations of Ling Zhang include Zhejiang University.

Papers
More filters
Proceedings ArticleDOI
22 Jul 2019
TL;DR: An inductance-based method is utilized to calculate the port priority fist, and afterwards deep reinforcement learning (DRL) with deep neural network (DNN) is applied to optimize the assignment of decaps on the prioritized locations.
Abstract: Selection of decoupling capacitors (decaps) is important for power distribution network (PDN) design in terms of lowering impedance and saving cost. Good PDN designs typically mean satisfying a target impedance with as less decaps as possible. In this paper, an inductance-based method is utilized to calculate the port priority fist, and afterwards deep reinforcement learning (DRL) with deep neural network (DNN) is applied to optimize the assignment of decaps on the prioritized locations. The DRL algorithm can explore by itself without any prior physical knowledge, and the DNN is trained with the exploration experience and eventually converges to an optimum state. The proposed hybrid method was tested on a printed-circuit-board (PCB) example. After some iterations of training the DNN successfully reached to an optimum design, which turned out to be the minimum number of decaps that can satisfy the target impedance. The usage of DRL with DNN makes the algorithm promising to include more variables as input and handle more complicated cases in the future.

17 citations

Journal ArticleDOI
TL;DR: The feasibility of sparse sampling is mathematically proved, and it is shown that increasing number of scanning points increases the signal-to-noise ratio of reconstructed images, and a nearest neighbor interpolation method is applied in the real-time processing to estimate the radiated power through the scanning plane.
Abstract: Emission source microscopy (ESM) technique can be utilized for the localization of electromagnetic interference sources in complex and large systems. This paper presents a sparse and nonuniform sampling technique for ESM. Compared with the traditional way of acquiring abundant and uniformly distributed scanning points on the scanning plane using a robotic scanning system, the proposed method is much more time-efficient in identifying the major radiation sources, even though the image quality is sacrificed. The feasibility of sparse sampling is mathematically proved, and it is shown that increasing number of scanning points increases the signal-to-noise ratio of reconstructed images. Besides, a nearest neighbor interpolation method is applied in the real-time processing to estimate the radiated power through the scanning plane. Thus, back-propagated images and estimated radiated power can be obtained in the real-time measurement process, which can efficiently and instantaneously provide the locations and the radiation strength of the most significant emission sources.

17 citations

Proceedings ArticleDOI
01 Jul 2020
TL;DR: This paper presents an improved decap-selection algorithm based on deep reinforcement learning (DRL), which seeks the minimum number of decaps through a self-exploration training to satisfy a given target impedance, and demonstrates the feasibility of achieving decent performance with pre-trained knowledge for more complicated engineering tasks in the future.
Abstract: The selection of decoupling capacitors (decap) is a critical but tedious process in power distribution network (PDN) design. In this paper, an improved decap-selection algorithm based on deep reinforcement learning (DRL), which seeks the minimum number of decaps through a self-exploration training to satisfy a given target impedance, is presented. Compared with the previous relevant work: the calculation speed of PDN impedance is significantly increased by adopting an impedance matrix reduction method; also, the enhanced algorithm performs a better convergence by utilizing the techniques of double Q-learning and prioritized experience replay; furthermore, a well-designed reward is proposed to facilitate long-term convergence when more decaps are required. The proposed algorithm demonstrates the feasibility of achieving decent performance using DRL with pre-trained knowledge for more complicated engineering tasks in the future.

15 citations

Journal ArticleDOI
TL;DR: The paper shows that sub-Nyquist is achievable and suggests uniform sampling is superior to nonuniform, in contrast to other reported uses of microwave imaging and care should be taken if the source reconstruction is based on uniform 2-D DFT.
Abstract: Sparse emission source microscopy (ESM) is an efficient method to identity radiating sources With the purpose to minimize the number of required measurement points, the presented work investigates how numerical properties of sparse ESM affects the quality of source reconstruction A simulation model of a simple printed circuit board was used instead of measurements to isolate the observed effect of the two-dimensional (2-D) discrete Fourier transformation (DFT) and the plane wave spectrum's numerical properties The paper shows that sub-Nyquist is achievable and suggests uniform sampling is superior to nonuniform, in contrast to other reported uses of microwave imaging Finally, the study shows that if the source reconstruction is based on uniform 2-D DFT care should be taken with the previously suggested intelligent selection of sparse samples based on real-time observation of the measured field

13 citations

Proceedings ArticleDOI
03 Jun 2019
TL;DR: The resulting I-V curve for the PN junction, using the deep learning solver presented in this work, shows a perfect match to the I-v curve obtained using the finite difference method, with the advantage of being 10 times faster at every time step.
Abstract: Simulating the dynamic characteristics of a PN junction at the microscopic level requires solving the Poisson's equation at every time step. Solving at every time step is a necessary but time-consuming process when using the traditional finite difference (FDM) approach. Deep learning is a powerful technique to fit complex functions. In this work, deep learning is utilized to accelerate solving Poisson's equation in a PN junction. The role of the boundary condition is emphasized in the loss function to ensure a better fitting. The resulting I-V curve for the PN junction, using the deep learning solver presented in this work, shows a perfect match to the I-V curve obtained using the finite difference method, with the advantage of being 10 times faster at every time step.

13 citations


Cited by
More filters
Proceedings ArticleDOI
R.W. Kelsall1
03 Apr 1995
TL;DR: If the authority ascribed to Monte Carlo models of devices at 1/spl mu/m feature size is to be maintained, modelling of the fundamental physics must be further improved, and the device model must be made more realistic.
Abstract: There can be little doubt that the Monte Carlo method for semiconductor device simulation has enormous power as a research tool. It represents a detailed physical model of the semiconductor material(s), and provides a high degree of insight into the microscopic transport processes. However, if the authority ascribed to Monte Carlo models of devices at 1/spl mu/m feature size is to be maintained for devices below O.1/spl mu/m, modelling of the fundamental physics must be further improved. And if the Monte Carlo method is to be successful as a semiconductor device design tool, the device model must be made more realistic. Success in the industrial sector depends on this, but also on achieving fast run-times optimisation - where the scope and need for ingenuity is now greatest.

436 citations

Proceedings Article
04 Jun 2019
TL;DR: In this paper, a deep neural network is proposed to learn a fast iterative solver tailored to a specific domain by learning to modify the updates of an existing solver using deep neural networks.
Abstract: Partial differential equations (PDEs) are widely used across the physical and computational sciences. Decades of research and engineering went into designing fast iterative solution methods. Existing solvers are general purpose, but may be sub-optimal for specific classes of problems. In contrast to existing hand-crafted solutions, we propose an approach to learn a fast iterative solver tailored to a specific domain. We achieve this goal by learning to modify the updates of an existing solver using a deep neural network. Crucially, our approach is proven to preserve strong correctness and convergence guarantees. After training on a single geometry, our model generalizes to a wide variety of geometries and boundary conditions, and achieves 2-3 times speedup compared to state-of-the-art solvers.

60 citations

Journal ArticleDOI
TL;DR: In this paper , the authors highlight some of the areas of highest potential impact, including to accelerate direct numerical simulations, to improve turbulence closure modeling, and to develop enhanced reduced-order models.
Abstract: Machine learning is rapidly becoming a core technology for scientific computing, with numerous opportunities to advance the field of computational fluid dynamics. In this Perspective, we highlight some of the areas of highest potential impact, including to accelerate direct numerical simulations, to improve turbulence closure modeling, and to develop enhanced reduced-order models. We also discuss emerging areas of machine learning that are promising for computational fluid dynamics, as well as some potential limitations that should be taken into account.

55 citations

Posted Content
TL;DR: This work proposes an approach to learn a fast iterative solver tailored to a specific domain by learning to modify the updates of an existing solver using a deep neural network, and achieves 2-3 times speedup compared to state-of-the-art solvers.
Abstract: Partial differential equations (PDEs) are widely used across the physical and computational sciences. Decades of research and engineering went into designing fast iterative solution methods. Existing solvers are general purpose, but may be sub-optimal for specific classes of problems. In contrast to existing hand-crafted solutions, we propose an approach to learn a fast iterative solver tailored to a specific domain. We achieve this goal by learning to modify the updates of an existing solver using a deep neural network. Crucially, our approach is proven to preserve strong correctness and convergence guarantees. After training on a single geometry, our model generalizes to a wide variety of geometries and boundary conditions, and achieves 2-3 times speedup compared to state-of-the-art solvers.

40 citations