Searcharxiv⌕ Search

arXiv subjects

Zhiwei Wang

Publications and source records attributed to Zhiwei Wang.

At least 217 records · Page 12Linked to original sources
arXiv2020

Siamese Neural Networks for Class Activity Detection

Classroom activity detection (CAD) aims at accurately recognizing speaker roles (either teacher or student) in classrooms. A CAD solution helps teachers get instant feedback on their pedagogical instructions. However, CAD is very challenging because (1) classroom conversations contain many conversational turn-taking overlaps between teachers and students; (2) the CAD model needs to be generalized well enough for different teachers and students; and (3) classroom recordings may be very noisy and low-quality. In this work, we address the above challenges by building a Siamese neural framework to automatically identify teacher and student utterances from classroom recordings. The proposed model is evaluated on real-world educational datasets. The results demonstrate that (1) our approach is superior on the prediction tasks for both online and offline classroom environments; and (2) our framework exhibits robustness and generalization ability on new teachers (i.e., teachers never appear in training data).

eess.AS↗
arXiv2020

Crystal structure and distortion of superconducting Cu$_x$Bi$_2$Se$_{3}$

The crystal structure of the candidate topological superconductor Cu$_x$Bi$_2$Se$_3$ was studied by single-crystal neutron diffraction using samples obtained by inserting the Cu dopant electrochemically. Neither structural refinements nor calculated scattering-density maps find a significant occupation of Cu at the intercalation site between the quintuple layers of Bi$_2$Se$_3$. Following Bragg reflection intensities as function of temperature, there is no signature of a structural phase transition between 295 and 2 K. However, the analysis of large sets of Bragg reflections indicates a small structural distortion breaking the rotational axis due to small displacements of the Bi ions.

cond-mat.supr-con↗
arXiv2020

Two-stage WECC Composite Load Modeling: A Double Deep Q-Learning Networks Approach

With the increasing complexity of modern power systems, conventional dynamic load modeling with ZIP and induction motors (ZIP + IM) is no longer adequate to address the current load characteristic transitions. In recent years, the WECC composite load model (WECC CLM) has shown to effectively capture the dynamic load responses over traditional load models in various stability studies and contingency analyses. However, a detailed WECC CLM model typically has a high degree of complexity, with over one hundred parameters, and no systematic approach to identifying and calibrating these parameters. Enabled by the wide deployment of PMUs and advanced deep learning algorithms, proposed here is a double deep Q-learning network (DDQN)-based, two-stage load modeling framework for the WECC CLM. This two-stage method decomposes the complicated WECC CLM for more efficient identification and does not require explicit model details. In the first stage, the DDQN agent determines an accurate load composition. In the second stage, the parameters of the WECC CLM are selected from a group of Monte-Carlo simulations. The set of selected load parameters is expected to best approximate the true transient responses. The proposed framework is verified using an IEEE 39-bus test system on commercial simulation platforms.

eess.SY↗
arXiv2020

Deriving AC OPF Solutions via Proximal Policy Optimization for Secure and Economic Grid Operation

Optimal power flow (OPF) is a very fundamental but vital optimization problem in the power system, which aims at solving a specific objective function (ex.: generator costs) while maintaining the system in the stable and safe operations. In this paper, we adopted the start-of-the-art artificial intelligence (AI) techniques to train an agent aiming at solving the AC OPF problem, where the nonlinear power balance equations are considered. The modified IEEE-14 bus system were utilized to validate the proposed approach. The testing results showed a great potential of adopting AI techniques in the power system operations.

eess.SY↗
arXiv2020

Extended Prony Analysis on Power System Oscillation Under a Near-Resonance Condition

Power system oscillations under a large disturbance often exhibit distorted waveforms as captured by increasingly deployed phasor measurement units. One cause is the occurrence of a near-resonance condition among several dominant modes that are influenced by nonlinear transient dynamics of generators. This paper proposes an Extended Prony Analysis method for measurement-based modal analysis. Based on the normal form theory, it compares analyses on transient and post-transient waveforms to distinguish a resonance mode caused by a near-resonance condition from natural modes so that the method can give more accurate modal properties than a traditional Prony Analysis method, especially for large disturbances. The new method is first demonstrated in detail on Kundur's two-area system and then tested on the IEEE 39-bus system to show its performance under a near-resonance condition.

eess.SY↗
arXiv2020

Implications of Stahl's Theorems to Holomorphic Embedding Pt. 2: Numerical Convergence

What has become known as Stahl's Theorem in power-engineering circles has been used to justify a convergence guarantee of the Holomorphic Embedding Method (HEM) as it applies to the power-flow problem. In this, the second part of a two-part paper, we examine implications to numerical convergence of HEM and the numerical properties of a Padé approximant algorithm. We show that even if the convergence domain is identical to the function's domain, numerical convergence of the sequence of Padé approximants computed with finite precision is not guaranteed. We also show that the study of convergence properties of the Padé approximant is the study of the location of branch-points of the function, which dictate branch-cut topology and capacity and, therefore, convergence rate. We show how poorly chosen embeddings can prevent numerical convergence.

eess.SY↗
arXiv2020

Implications of Stahl's Theorems to Holomorphic Embedding Pt. 1: Theoretical Convergence

What has become known as Stahl's Theorem in power engineering circles has been used to justify a convergence guarantee of the Holormorphic Embedding Method (HEM) as it applies to the power flow (PF) problem. In this two-part paper, we examine in more detail the implications of Stahl's theorems to both theoretcial and numerical convergence for a wider range of problems to which these theorems are now being applied. In Pt. 1, we introduce the theorem using the necessary mathematical parlance and then translate the language to show its implications to convergence of nonlinear problems in general and the PF problem specifically. We show that among other possibilities the existence of the Chebotarev points, which are embedding specific, are a possible theoretical impediment to convergence. Numerical impediments to convergences are discussed in the companion paper.

eess.SY↗
arXiv2020

Averages of coefficients of a class of degree 3 L-functions

In this note, we give a detailed proof of an asymptotic for averages of coefficients of a class of degree three $L$-functions which can be factorized as a product of a degree one and a degree two $L$-functions. We emphasize that we can break the $1/2$-barrier in the error term, and we get an explicit exponent.

math.NT↗
arXiv2020

Distributed Frequency Emergency Control with Coordinated Edge Intelligence

Developing effective strategies to rapidly support grid frequency while minimizing loss in case of severe contingencies is an important requirement in power systems. While distributed responsive load demands are commonly adopted for frequency regulation, it is difficult to achieve both rapid response and global accuracy in a practical and cost-effective manner. In this paper, the cyber-physical design of an Internet-of-Things (IoT) enabled system, called Grid Sense, is presented. Grid Sense utilizes a large number of distributed appliances for frequency emergency support. It features a local power loss $ΔP$ estimation approach for frequency emergency control based on coordinated edge intelligence. The specifically designed smart outlets of Grid Sense detect the frequency disturbance event locally using the parameters sent from the control center to estimate active power loss in the system and to make rapid and accurate switching decisions soon after a severe contingency. Based on a modified IEEE 24-bus system, numerical simulations and hardware experiments are conducted to demonstrate the frequency support performance of Grid Sense in the aspects of accuracy and speed. It is shown that Grid Sense equipped with its local $ΔP$-estimation frequency control approach can accurately and rapidly prevent the drop of frequency after a major power loss.

eess.SY↗
arXiv2020

Graph Computing based Distributed State Estimation with PMUs

Power system state estimation plays a fundamental and critical role in the energy management system (EMS). To achieve a high performance and accurate system states estimation, a graph computing based distributed state estimation approach is proposed in this paper. Firstly, a power system network is divided into multiple areas. Reference buses are selected with PMUs being installed at these buses for each area. Then, the system network is converted into multiple independent areas. In this way, the power system state estimation could be conducted in parallel for each area and the estimated system states are obtained without compromise of accuracy. IEEE 118-bus system and MP 10790-bus system are employed to verify the results accuracy and present the promising computation performance.

cs.DC↗
arXiv2020

Online Low Frequency Oscillation Detection and Analysis System with an Ensemble Filter

The widespread deployment of phasor measurement unit (PMU) overpower systems makes it possible to monitor and analyze grid dynamics in real-time. Low-frequency oscillation is harmful to power system equipment and operation, and in the worst-case scenario may lead to cascading failures. Therefore, it is critical to detect and identify them as soon as they appear. This paper presents an online low-frequency oscillation detection and analysis (LFODA) system, which has the merit of significantly reducing the chance of false alarm via a voting schema and a time-serial filter. A novel algorithm based on density-based spatial clustering of applications with noise (DBSCAN) is proposed to classify oscillation modes as well as to group their corresponding buses/monitoring sites. Performance of the LFODA system is evaluated through experiments using both simulated and real-world PMU data.

eess.SP↗
arXiv2020

Positivity of holomorphic vector bundles in terms of $L^p$-conditions of $\bar\partial$

We study the positivity properties of Hermitian (or even Finsler) holomorphic vector bundles in terms of $L^p$-estimates of $\bar\partial$ and $L^p$-extensions of holomorphic objects. To this end, we introduce four conditions, called the optimal $L^p$-estimate condition, the multiple coarse $L^p$-estimate condition, the optimal $L^p$-extension condition, and the multiple coarse $L^p$-extension condition, for a Hermitian (or Finsler) vector bundle $(E,h)$. The main result of the present paper is to give a characterization of the Nakano positivity of $(E,h)$ via the optimal $L^2$-estimate condition. We also show that $(E,h)$ is Griffiths positive if it satisfies the multiple coarse $L^p$-estimate condition for some $p>1$, the optimal $L^p$-extension condition, or the multiple coarse $L^p$-extension condition for some $p>0$. These results can be roughly viewed as converses of Hörmander's $L^2$-estimate of $\bar\partial$ and Ohsawa-Takegoshi type extension theorems. As an application of the main result, we get a totally different method to Nakano positivity of direct image sheaves of twisted relative canonical bundles associated to holomorphic families of complex manifolds.

math.CV↗
arXiv2019

Label Refinement with an Iterative Generative Adversarial Network for Boosting Retinal Vessel Segmentation

State-of-the-art methods for retinal vessel segmentation mainly rely on manually labeled vessels as the ground truth for supervised training. The quality of manual labels plays an essential role in the segmentation accuracy, while in practice it could vary a lot and in turn could substantially mislead the training process and limit the segmentation accuracy. This paper aims to "purify" any comprehensive training set, which consists of data annotated by various observers, via refining low-quality manual labels in the dataset. To this end, we have developed a novel label refinement method based on an iterative generative adversarial network (GAN). Our iterative GAN is trained based on a set of high-quality patches (i.e. with consistent manual labels among different observers) and low-quality patches with noisy manual vessel labels. A simple yet effective method has been designed to simulate low-quality patches with noises which conform to the distribution of real noises from human observers. To evaluate the effectiveness of our method, we have trained four state-of-the-art retinal vessel segmentation models using the purified dataset obtained from our method and compared their performance with those trained based on the original noisy datasets. Experimental results on two datasets DRIVE and CHASE_DB1 demonstrate that obvious accuracy improvements can be achieved for all the four models when using the purified datasets from our method.

eess.IV↗
arXiv2019

Learning Multi-level Dependencies for Robust Word Recognition

Robust language processing systems are becoming increasingly important given the recent awareness of dangerous situations where brittle machine learning models can be easily broken with the presence of noises. In this paper, we introduce a robust word recognition framework that captures multi-level sequential dependencies in noised sentences. The proposed framework employs a sequence-to-sequence model over characters of each word, whose output is given to a word-level bi-directional recurrent neural network. We conduct extensive experiments to verify the effectiveness of the framework. The results show that the proposed framework outperforms state-of-the-art methods by a large margin and they also suggest that character-level dependencies can play an important role in word recognition.

cs.CL↗
arXiv2019

AI-Based Autonomous Line Flow Control via Topology Adjustment for Maximizing Time-Series ATCs

This paper presents a novel AI-based approach for maximizing time-series available transfer capabilities (ATCs) via autonomous topology control considering various practical constraints and uncertainties. Several AI techniques including supervised learning and deep reinforcement learning (DRL) are adopted and improved to train effective AI agents for achieving the desired performance. First, imitation learning (IL) is used to provide a good initial policy for the AI agent. Then, the agent is trained by DRL algorithms with a novel guided exploration technique, which significantly improves the training efficiency. Finally, an Early Warning (EW) mechanism is designed to help the agent find good topology control strategies for long testing periods, which helps the agent to determine action timing using power system domain knowledge; thus, effectively increases the system error-tolerance and robustness. Effectiveness of the proposed approach is demonstrated in the "2019 Learn to Run a Power Network (L2RPN)" global competition, where the developed AI agents can continuously and safely control a power grid to maximize ATCs without operator's intervention for up to 1-month's operation data and eventually won the first place in both development and final phases of the competition. The winning agent has been open-sourced on GitHub.

eess.SP↗
arXiv2019

Community Microgrid Planning Considering Building Thermal Dynamics

In this paper, the building thermal dynamic characteristics are introduced in the community microgrid (MG) planning model. The proposed planning model is formulated as a mixed integer linear programming (MILP) which seeks to determine the optimal deployment strategy for various distributed energy resources (DER). The objective is to minimize the annualized cost of community MG, including investment cost in DER, operation cost for dispatchable fuel-generators (DFG), energy storage system (ESS) degradation cost, energy purchasing and peak demand charge at PCC, customer discomfort cost due to the room temperature deviation and load curtailment cost. Given the slow thermal dynamic characteristics of buildings, the heating, ventilation and air-conditioning (HVAC) system in the proposed model is treated as a demand side management (DSM) component, whose dispatch commands are provided by the central MG controller. Numerical results based on a community MG comprising of 20 residential buildings demonstrate the effectiveness of the proposed planning model and the benefits of introducing the building thermal dynamic model.

math.OC↗
arXiv2019

Uniaxial-strain Control of Nematic Superconductivity in Sr$_{x}$Bi$_2$Se$_3$

Nematic states are characterized by rotational symmetry breaking without translational ordering. Recently, nematic superconductivity, in which the superconducting gap spontaneously lifts the rotational symmetry of the lattice, has been discovered. However the pairing mechanism and the mechanism determining the nematic orientation remain unresolved. A first step is to demonstrate control of the nematicity, through application of an external symmetry-breaking field, to determine the sign and strength of coupling to the lattice. Here, we report for the first time control of the nematic orientation of the superconductivity of Sr$_x$Bi$_2$Se$_3$, through externally-applied uniaxial stress. The suppression of subdomains indicates that it is the $Δ_{4y}$ state that is most favoured under compression along the basal Bi-Bi bonds. These results provide an inevitable step towards understanding the microscopic origin of the unique topological nematic superconductivity.

cond-mat.supr-con↗