Showing papers by "École Polytechnique de Montréal published in 2015"

BinaryConnect: training deep neural networks with binary weights during propagations

Abstract: Deep Neural Networks (DNN) have achieved state-of-the-art results in a wide range of tasks, with the best results obtained with large training sets and large models. In the past, GPUs enabled these breakthroughs because of their greater computational speed. In the future, faster computation at both training and test time is likely to be crucial for further progress and for consumer applications on low-power devices. As a result, there is much interest in research and development of dedicated hardware for Deep Learning (DL). Binary weights, i.e., weights which are constrained to only two possible values (e.g. -1 or 1), would bring great benefits to specialized DL hardware by replacing many multiply-accumulate operations by simple accumulations, as multipliers are the most space and power-hungry components of the digital implementation of neural networks. We introduce BinaryConnect, a method which consists in training a DNN with binary weights during the forward and backward propagations, while retaining precision of the stored weights in which gradients are accumulated. Like other dropout schemes, we show that BinaryConnect acts as regularizer and we obtain near state-of-the-art results with BinaryConnect on the permutation-invariant MNIST, CIFAR-10 and SVHN.

Photooxidation and quantum confinement effects in exfoliated black phosphorus

Abstract: The degradation of exfoliated black phosphorus in ambient conditions may limit its use in electronic devices. The combined effects of light irradiation and exposure to oxygen on mono- and multilayers of this material are now investigated.

Describing Videos by Exploiting Temporal Structure

Abstract: Recent progress in using recurrent neural networks (RNNs) for image description has motivated the exploration of their application for video description. However, while images are static, working with videos requires modeling their dynamic temporal structure and then properly integrating that information into a natural language description model. In this context, we propose an approach that successfully takes into account both the local and global temporal structure of videos to produce descriptions. First, our approach incorporates a spatial temporal 3-D convolutional neural network (3-D CNN) representation of the short temporal dynamics. The 3-D CNN representation is trained on video action recognition tasks, so as to produce a representation that is tuned to human motion and behavior. Second we propose a temporal attention mechanism that allows to go beyond local temporal modeling and learns to automatically select the most relevant temporal segments given the text-generating RNN. Our approach exceeds the current state-of-art for both BLEU and METEOR metrics on the Youtube2Text dataset. We also present results on a new, larger and more challenging dataset of paired video and natural language descriptions.

Strong light-matter coupling in two-dimensional atomic crystals

Abstract: Microcavity polaritons—the bosonic quasiparticles that result from strong light–matter coupling—are observed for the first time in a dielectric cavity containing a monolayer of molybdenum disulphide at room temperature.

BinaryConnect: Training Deep Neural Networks with binary weights during propagations

Abstract: Deep Neural Networks (DNN) have achieved state-of-the-art results in a wide range of tasks, with the best results obtained with large training sets and large models. In the past, GPUs enabled these breakthroughs because of their greater computational speed. In the future, faster computation at both training and test time is likely to be crucial for further progress and for consumer applications on low-power devices. As a result, there is much interest in research and development of dedicated hardware for Deep Learning (DL). Binary weights, i.e., weights which are constrained to only two possible values (e.g. -1 or 1), would bring great benefits to specialized DL hardware by replacing many multiply-accumulate operations by simple accumulations, as multipliers are the most space and power-hungry components of the digital implementation of neural networks. We introduce BinaryConnect, a method which consists in training a DNN with binary weights during the forward and backward propagations, while retaining precision of the stored weights in which gradients are accumulated. Like other dropout schemes, we show that BinaryConnect acts as regularizer and we obtain near state-of-the-art results with BinaryConnect on the permutation-invariant MNIST, CIFAR-10 and SVHN.

SuBSENSE: A Universal Change Detection Method With Local Adaptive Sensitivity

Abstract: Foreground/background segmentation via change detection in video sequences is often used as a stepping stone in high-level analytics and applications. Despite the wide variety of methods that have been proposed for this problem, none has been able to fully address the complex nature of dynamic scenes in real surveillance tasks. In this paper, we present a universal pixel-level segmentation method that relies on spatiotemporal binary features as well as color information to detect changes. This allows camouflaged foreground objects to be detected more easily while most illumination variations are ignored. Besides, instead of using manually set, frame-wide constants to dictate model sensitivity and adaptation speed, we use pixel-level feedback loops to dynamically adjust our method’s internal parameters without user intervention. These adjustments are based on the continuous monitoring of model fidelity and local segmentation noise levels. This new approach enables us to outperform all 32 previously tested state-of-the-art methods on the 2012 and 2014 versions of the ChangeDetection.net dataset in terms of overall F-Measure. The use of local binary image descriptors for pixel-level modeling also facilitates high-speed parallel implementations: our own version, which used no low-level or architecture-specific instruction, reached real-time processing speed on a midlevel desktop CPU. A complete C++ implementation based on OpenCV is available online.

Delving Deeper into Convolutional Networks for Learning Video Representations

Abstract: We propose an approach to learn spatio-temporal features in videos from intermediate visual representations we call "percepts" using Gated-Recurrent-Unit Recurrent Networks (GRUs).Our method relies on percepts that are extracted from all level of a deep convolutional network trained on the large ImageNet dataset. While high-level percepts contain highly discriminative information, they tend to have a low-spatial resolution. Low-level percepts, on the other hand, preserve a higher spatial resolution from which we can model finer motion patterns. Using low-level percepts can leads to high-dimensionality video representations. To mitigate this effect and control the model number of parameters, we introduce a variant of the GRU model that leverages the convolution operations to enforce sparse connectivity of the model units and share parameters across the input spatial locations. We empirically validate our approach on both Human Action Recognition and Video Captioning tasks. In particular, we achieve results equivalent to state-of-art on the YouTube2Text dataset using a simpler text-decoder model and without extra 3D CNN features.

Intraoperative brain cancer detection with Raman spectroscopy in humans

Abstract: Cancers are often impossible to visually distinguish from normal tissue. This is critical for brain cancer where residual invasive cancer cells frequently remain after surgery, leading to disease recurrence and a negative impact on overall survival. No preoperative or intraoperative technology exists to identify all cancer cells that have invaded normal brain. To address this problem, we developed a handheld contact Raman spectroscopy probe technique for live, local detection of cancer cells in the human brain. Using this probe intraoperatively, we were able to accurately differentiate normal brain from dense cancer and normal brain invaded by cancer cells, with a sensitivity of 93% and a specificity of 91%. This Raman-based probe enabled detection of the previously undetectable diffusely invasive brain cancer cells at cellular resolution in patients with grade 2 to 4 gliomas. This intraoperative technology may therefore be able to classify cell populations in real time, making it an ideal guide for surgical resection and decision-making.

Melanins and melanogenesis: from pigment cells to human health and technological applications.

Abstract: During the past decade, melanins and melanogenesis have attracted growing interest for a broad range of biomedical and technological applications. The burst of polydopamine-based multifunctional coatings in materials science is just one example, and the list may be expanded to include melanin thin films for organic electronics and bioelectronics, drug delivery systems, functional nanoparticles and biointerfaces, sunscreens, environmental remediation devices. Despite considerable advances, applied research on melanins and melanogenesis is still far from being mature. A closer intersectoral interaction between research centers is essential to raise the interests and increase the awareness of the biomedical, biomaterials science and hi-tech sectors of the manifold opportunities offered by pigment cells and related metabolic pathways. Starting from a survey of biological roles and functions, the present review aims at providing an interdisciplinary perspective of melanin pigments and related pathway with a view to showing how it is possible to translate current knowledge about physical and chemical properties and control mechanisms into new bioinspired solutions for biomedical, dermocosmetic, and technological applications.

Recurrent Neural Networks for Emotion Recognition in Video

Abstract: Deep learning based approaches to facial analysis and video analysis have recently demonstrated high performance on a variety of key tasks such as face recognition, emotion recognition and activity recognition. In the case of video, information often must be aggregated across a variable length sequence of frames to produce a classification result. Prior work using convolutional neural networks (CNNs) for emotion recognition in video has relied on temporal averaging and pooling operations reminiscent of widely used approaches for the spatial aggregation of information. Recurrent neural networks (RNNs) have seen an explosion of recent interest as they yield state-of-the-art performance on a variety of sequence analysis tasks. RNNs provide an attractive framework for propagating information over a sequence using a continuous valued hidden layer representation. In this work we present a complete system for the 2015 Emotion Recognition in the Wild (EmotiW) Challenge. We focus our presentation and experimental analysis on a hybrid CNN-RNN architecture for facial expression analysis that can outperform a previously applied CNN approach using temporal averaging for aggregation.

General Metasurface Synthesis Based on Susceptibility Tensors

Abstract: A general method, based on susceptibility tensors, is proposed for the synthesis of metasurfaces transforming arbitrary incident waves into arbitrary reflected and transmitted waves. The proposed method exhibits two advantages: 1) it is inherently vectorial, and therefore better suited for full vectorial (beyond paraxial) electromagnetic problems; 2) it provides closed-form solutions, and is therefore extremely fast. Incidentally, the method reveals that a metasurface is fundamentally capable to transform up to four independent wave triplets (incident, reflected, and refracted waves). In addition, this paper provides the closed-form expressions relating the synthesized susceptibilities and the scattering parameters simulated within periodic boundary conditions, which allows one to design the scattering particles realizing the desired susceptibilities. The versatility of the method is illustrated by examples of metasurfaces achieving the following transformations: generalized refraction, reciprocal and nonreciprocal polarization rotation, Bessel vortex beam generation, and orbital angular momentum multiplexing.

Importance of proximity to resources, social support, transportation and neighborhood security for mobility and social participation in older adults: results from a scoping study

Abstract: Since mobility and social participation are key determinants of health and quality of life, it is important to identify factors associated with them. Although several investigations have been conducted on the neighborhood environment, mobility and social participation, there is no clear integration of the results. This study aimed to provide a comprehensive understanding regarding how the neighborhood environment is associated with mobility and social participation in older adults. A rigorous methodological scoping study framework was used to search nine databases from different fields with fifty-one keywords. Data were exhaustively analyzed, organized and synthesized according to the International Classification of Functioning, Disability and Health (ICF) by two research assistants following PRISMA guidelines, and results were validated with knowledge users. The majority of the 50 selected articles report results of cross-sectional studies (29; 58 %), mainly conducted in the US (24; 48 %) or Canada (15; 30 %). Studies mostly focused on neighborhood environment associations with mobility (39; 78 %), social participation (19; 38 %), and occasionally both (11; 22 %). Neighborhood attributes considered were mainly ‘Pro ducts and technology’ (43; 86 ) and ‘Services, systems and policies’ (37; 74 %), but also ‘Natural and human-made changes’ (27; 54 %) and ‘Support and relationships’ (21; 42 %). Mobility and social participation were both positively associated with Proximity to resources and recreational facilities, Social support, Having a car or driver’s license, Public transportation and Neighborhood security, and negatively associated with Poor user-friendliness of the walking environment and Neighborhood insecurity. Attributes of the neighborhood environment not covered by previous research on mobility and social participation mainly concerned ‘Attitudes’, and ‘Services, systems and policies’. Results from this comprehensive synthesis of empirical studies on associations of the neighborhood environment with mobility and social participation will ultimately support best practices, decisions and the development of innovative inclusive public health interventions including clear guidelines for the creation of age-supportive environments. To foster mobility and social participation, these interventions must consider Proximity to resources and to recreational facilities, Social support, Transportation, Neighborhood security and User-friendliness of the walking environment. Future studies should include both mobility and social participation, and investigate how they are associated with ‘Attitudes’, and ‘Services, systems and policies’ in older adults, including disadvantaged older adults.

Describing Videos by Exploiting Temporal Structure

Abstract: Recent progress in using recurrent neural networks (RNNs) for image description has motivated the exploration of their application for video description. However, while images are static, working with videos requires modeling their dynamic temporal structure and then properly integrating that information into a natural language description. In this context, we propose an approach that successfully takes into account both the local and global temporal structure of videos to produce descriptions. First, our approach incorporates a spatial temporal 3-D convolutional neural network (3-D CNN) representation of the short temporal dynamics. The 3-D CNN representation is trained on video action recognition tasks, so as to produce a representation that is tuned to human motion and behavior. Second we propose a temporal attention mechanism that allows to go beyond local temporal modeling and learns to automatically select the most relevant temporal segments given the text-generating RNN. Our approach exceeds the current state-of-art for both BLEU and METEOR metrics on the Youtube2Text dataset. We also present results on a new, larger and more challenging dataset of paired video and natural language descriptions.

Air-Filled Substrate Integrated Waveguide for Low-Loss and High Power-Handling Millimeter-Wave Substrate Integrated Circuits

Abstract: An air-filled substrate integrated waveguide (SIW) made of a multilayer printed circuit board process is proposed in this paper. It is of particular interest for millimeter-wave applications that generally require low cost and low-loss performance and excellent power-handling capability. This three-layered air-filled SIW allows for substantial loss reduction and power-handling capability enhancement. The top and bottom layers may make use of a low-cost standard substrate such as FR-4 on which baseband or digital circuits can be implemented so to obtain a very compact, high-performance, low-cost, and self-packaged millimeter-wave integrated system. Over Ka-band (U-band), it is shown that the air-filled SIW compared to its dielectric-filled counterparts based on Rogers substrates RT/Duroid 5880 and also 6002 reduces losses by a mean value of 0.068 dB/cm (0.098 dB/cm) and 0.104 dB/cm (0.152 dB/cm), increases average power-handling capability by 8 dB (6 dB) and 7.5 dB (5.7 dB), and quality factor by 2.7 (2.8) and 3.6 (3.8) times, respectively. The peak power-handling capability of the proposed structure is also studied. A wideband transition is presented to facilitate interconnects of the proposed air-filled SIW with dielectric-filled SIW. Design steps of this transition are detailed and its bandwidth limitation due to fabrication tolerances is theoretically examined and established. For validation purposes, a back-to-back transition operating over the Ka-band is fabricated. It achieves a return loss of better than 15 dB and an insertion loss of ${\hbox{0.6}} \pm {\hbox{0.2 dB}}$ ( ${\hbox{0.3}} \pm {\hbox{0.1}}~{\hbox{dB}}$ for the transition) from 27 to 40 GHz. Finally, two elementary circuits, namely, the T-junction and 90 $^{\circ}$ hybrid coupler based on the air-filled SIW, are also demonstrated.

Quantifying the microvascular origin of BOLD-fMRI from first principles with two-photon microscopy and an oxygen-sensitive nanoprobe.

Abstract: The blood oxygenation level-dependent (BOLD) contrast is widely used in functional magnetic resonance imaging (fMRI) studies aimed at investigating neuronal activity. However, the BOLD signal reflects changes in blood volume and oxygenation rather than neuronal activity per se. Therefore, understanding the transformation of microscopic vascular behavior into macroscopic BOLD signals is at the foundation of physiologically informed noninvasive neuroimaging. Here, we use oxygen-sensitive two-photon microscopy to measure the BOLD-relevant microvascular physiology occurring within a typical rodent fMRI voxel and predict the BOLD signal from first principles using those measurements. The predictive power of the approach is illustrated by quantifying variations in the BOLD signal induced by the morphological folding of the human cortex. This framework is then used to quantify the contribution of individual vascular compartments and other factors to the BOLD signal for different magnet strengths and pulse sequences.

History and Applications of Hydrogels

Abstract: Hydrogels have existed for more than half century, providing one of the earliest records of crosslinked hydroxyethyl methacrylate (HEMA) hydrogels. Today, hydrogels still fascinate material scientists and biomedical researchers and great strides have been made in terms of their formulations and applications. As a class of material, hydrogels are unique, they consist of a self-supporting, water-swollen three-dimensional (3D) viscoelastic network which permits the diffusion and attachment of molecules and cells. However, hydrogels have recently drawn great attention for use in a wide variety of biomedical applications such as cell therapeutics, wound healing, cartilage/bone regeneration and the sustained release of drugs. This is due to their biocompatibility and the similarity of their physical properties to natural tissue. This review aims to give an overview of the historic and the recent design concept of hydrogels and their several applications based on the old and the most recent publications in this field.

Direct oriented growth of armchair graphene nanoribbons on germanium.

Abstract: Graphene can be transformed from a semimetal into a semiconductor if it is confined into nanoribbons narrower than 10 nm with controlled crystallographic orientation and well-defined armchair edges. However, the scalable synthesis of nanoribbons with this precision directly on insulating or semiconducting substrates has not been possible. Here we demonstrate the synthesis of graphene nanoribbons on Ge(001) via chemical vapour deposition. The nanoribbons are self-aligning 3° from the Ge〈110〉 directions, are self-defining with predominantly smooth armchair edges, and have tunable width to 70. In order to realize highly anisotropic ribbons, it is critical to operate in a regime in which the growth rate in the width direction is especially slow, <5 nm h(-1). This directional and anisotropic growth enables nanoribbon fabrication directly on conventional semiconductor wafer platforms and, therefore, promises to allow the integration of nanoribbons into future hybrid integrated circuits.

Are bullies more productive?: empirical study of affectiveness vs. issue fixing time

Abstract: Human Affectiveness, i.e., the emotional state of a person, plays a crucial role in many domains where it can make or break a team's ability to produce successful products. Software development is a collaborative activity as well, yet there is little information on how affectiveness impacts software productivity. As a first measure of this impact, this paper analyzes the relation between sentiment, emotions and politeness of developers in more than 560K Jira comments with the time to fix a Jira issue. We found that the happier developers are (expressing emotions such as JOY and LOVE in their comments), the shorter the issue fixing time is likely to be. In contrast, negative emotions such as SADNESS, are linked with longer issue fixing time. Politeness plays a more complex role and we empirically analyze its impact on developers' productivity.

A systematic literature review on the usage of eye-tracking in software engineering

Abstract: ContextEye-tracking is a mean to collect evidence regarding some participants' cognitive processes. Eye-trackers monitor participants' visual attention by collecting eye-movement data. These data are useful to get insights into participants' cognitive processes during reasoning tasks. ObjectiveThe Evidence-based Software Engineering (EBSE) paradigm has been proposed in 2004 and, since then, has been used to provide detailed insights regarding different topics in software engineering research and practice. Systematic Literature Reviews (SLR) are also useful in the context of EBSE by bringing together all existing evidence of research and results about a particular topic. This SLR evaluates the current state of the art of using eye-trackers in software engineering and provides evidence on the uses and contributions of eye-trackers to empirical studies in software engineering. MethodWe perform a SLR covering eye-tracking studies in software engineering published from 1990 up to the end of 2014. To search all recognised resources, instead of applying manual search, we perform an extensive automated search using Engineering Village. We identify 36 relevant publications, including nine journal papers, two workshop papers, and 25 conference papers. ResultsThe software engineering community started using eye-trackers in the 1990s and they have become increasingly recognised as useful tools to conduct empirical studies from 2006. We observe that researchers use eye-trackers to study model comprehension, code comprehension, debugging, collaborative interaction, and traceability. Moreover, we find that studies use different metrics based on eye-movement data to obtain quantitative measures. We also report the limitations of current eye-tracking technology, which threaten the validity of previous studies, along with suggestions to mitigate these limitations. ConclusionHowever, not withstanding these limitations and threats, we conclude that the advent of new eye-trackers makes the use of these tools easier and less obtrusive and that the software engineering community could benefit more from this technology.

Higher levels of cardiovascular fitness are associated with better executive function and prefrontal oxygenation in younger and older women.

Abstract: Aim: Many studies have suggested that physical exercise training improves cognition and more selectively executive functions. There is a growing interest to clarify the neurophysiological mechanisms that underlie this effect. The aim of the current study was to evaluate the neurophysiological changes in cerebral oxygenation associated with physical fitness level and executive functions. Method: In this study, 22 younger and 36 older women underwent a maximal graded continuous test (i.e., O2max) in order to classifyassign them into a fitness group (higher vs. lower fit). All participants completed neuropsychological paper and pencil testing and a computerized Stroop task (which contained executive and non-executive conditions) in which the change in pPrefrontal cortex oxygenation change was evaluated in all participants with a near infrared spectroscopy (NIRS). system during a computerized Stroop task (which contains executive and non-executive conditions). Results: Our findings revealed a Fitness x Condition interaction (p < .05) such that higher fit women scored better on measures of executive functions than lower fit women. In comparison to lower fit women, higher fit women had faster reaction times in the switching (executive)Executive condition of the computerized Stroop task. No significant effect was observed ion the non-executive condition of the test and no interactions were found with age. In measures of cerebral oxygenation (ΔHbT and ΔHbO2), we found a main effect of fitness on cerebral oxygenation during the Stroop task such that only high fit women demonstrated a significant increase in the right inferior frontal gyrus. Discussion/Conclusion: Higher fit individuals who demonstrate better cardiorespiratory functions (as measured by O2max) show faster reaction times and greater cerebral oxygenation in the right inferior frontal gyrus than women with lower fitness levels. The lack of interaction with age, suggests that good cardiorespiratory functions

Decentralized active information acquisition: Theory and application to multi-robot SLAM

Abstract: This paper addresses the problem of controlling mobile sensing systems to improve the accuracy and efficiency of gathering information autonomously. It applies to scenarios such as environmental monitoring, search and rescue, surveillance and reconnaissance, and simultaneous localization and mapping (SLAM). A multi-sensor active information acquisition problem, capturing the common characteristics of these scenarios, is formulated. The goal is to design sensor control policies which minimize the entropy of the estimation task, conditioned on the future measurements. First, we provide a non-greedy centralized solution, which is computationally fast, since it exploits linearized sensing models, and memory efficient, since it exploits sparsity in the environment model. Next, we decentralize the control task to obtain linear complexity in the number of sensors and provide suboptimality guarantees. Finally, our algorithms are applied to the multi-robot active SLAM problem to enable a decentralized nonmyopic solution that exploits sparsity in the planning process.

Comparative grading scales, statistical analyses, climber descriptors and ability grouping: International Rock Climbing Research Association position statement

Abstract: The research base for rock climbing has expanded substantially in the past three decades as worldwide interest in the sport has grown. An important trigger for the increasing research attention has been the transition of the sport to a competitive as well as recreational activity and the potential inclusion of sport climbing in the Olympic schedule. The International Rock Climbing Research Association (IRCRA) was formed in 2011 to bring together climbers, coaches and researchers to share knowledge and promote collaboration. This position statement was developed during and after the 2nd IRCRA Congress which was held in Pontresina, in September 2014. The aim of the position statement is to bring greater uniformity to the descriptive and statistical methods used in reporting rock climbing research findings. To date there is a wide variation in the information provided by researchers regarding the climbers’ characteristics and also in the approaches employed to convert from climbing grading scales to...

Protonic and Electronic Transport in Hydrated Thin Films of the Pigment Eumelanin

Abstract: The electrical properties of eumelanin, a ubiquitous natural pigment, have fascinated scientists since the late 1960s. For several decades, the hydration-dependent electrical properties of eumelanin have mainly been interpreted within the amorphous semiconductor model. Recent works undermined this paradigm. Here we study protonic and electronic charge carrier transport in hydrated eumelanin in thin film form. Thin films are ideal candidates for these studies since they are readily accessible to chemical and morphological characterization and potentially amenable to device applications. Current–voltage (I-V) measurements, transient current measurements with proton-transparent electrodes, and electrochemical impedance spectroscopy (EIS) measurements are reported and correlated with the results of the chemical characterization of the films, performed by X-ray photoelectron spectroscopy. We show that the electrical response of hydrated eumelanin films is dominated by ionic conduction (10–4–10–3 S cm–1), large...

Solvent-induced changes in PEDOT:PSS films for organic electrochemical transistors

Abstract: Organic electrochemical transistors based on the conducting polymer poly(3,4-ethylenedioxythiophene) doped with poly(styrenesulfonate) (PEDOT:PSS) are of interest for several bioelectronic applications In this letter, we investigate the changes induced by immersion of PEDOT:PSS films, processed by spin coating from different mixtures, in water and other solvents of different polarities We found that the film thickness decreases upon immersion in polar solvents, while the electrical conductivity remains unchanged The decrease in film thickness is minimized via the addition of a cross-linking agent to the mixture used for the spin coating of the films

Potential and limits of numerical modelling for supporting the development of HTS devices

Abstract: In this paper, we present a general review of the status of numerical modelling applied to the design of high temperature superconductor devices. The importance of this tool is emphasized at the beginning of the paper, followed by formal definitions of the notions of models, numerical methods and numerical models. The state-of-the-art models are listed, and the main limitations of existing numerical models are reported. Those limitations are shown to concern two aspects: on the one hand, the numerical performance (i.e. speed) of the methods themselves is not good enough yet; on the other hand, the availability of model file templates, material data and benchmark problems is clearly insufficient. Paths for improving those elements are indicated in the paper. Besides the technical aspects of the research to be further pursued, for instance in adaptive numerical methods, most recommendations command for an increased collective effort for sharing files, data, codes and their documentation.

Real-Time Simulation of MMCs Using CPU and FPGA

Abstract: Modular multilevel converter (MMC) structures are composed of several hundreds to thousands of half-bridge converters. Such large numbers of power switches and electrical nodes introduce important numerical challenges for the computation of electromagnetic transients. The problem becomes particularly more complex for the real-time simulations. This paper presents a feasibility study on the real-time simulation of the MMC models. CPU-based and field-programmable gate array-based implementations are proposed and evaluated for the MMCs having up to 401 levels. The study also provides guidelines for the real-time simulation platform requirements to simulate these MMC models.