Showing papers by "ParisTech published in 2018"

The Migrate project: the challenges of operating a transmission grid with only inverter-based generation. A grid-forming control improvement with transient current-limiting control

Abstract: Renewable generation is mainly connected through converters. Even if they provide more and more ancillary services to the grid, these may not be sufficient for extremely high penetrations. As the share of such generating units is growing rapidly, some synchronous areas could in the future occasionally be operated without synchronous machines. In such conditions, system behaviour will dramatically change, but stability will still have to be ensured with the same level of reliability as today. To reach this ambitious goal, the control of inverters will have to be changed radically. Inverters will need to move from following the grid to leading the grid behaviour, both in steady state and during transients. This new type of control brings additional issues on converters that are addressed in this study. A solution is proposed to allow a stable operation of the system together with a limited solicitation of inverters during transients.

Tuning of Cascaded Controllers for Robust Grid-Forming Voltage Source Converter

Abstract: From the origin of the grid, energy has been delivered to electrical loads mainly by synchronous machines. All the main rules to manage the grid have been based on the electromechanical behavior of these machines which have been extensively studied for many years. Due to the increase of HVDC link and renewable energy sources as wind turbine and PV, power converters are massively introduced in the grid with a fundamentally different dynamic behavior. Some years ago, they were connected as simple power injector. Then, they were asked to provide some ancillary services to the grid, in the future, grid forming capability will be required. Even if grid-forming converters had been extensively studied for microgrids and offshore grids, it has to be adapted to transmission grid where the topology may be largely modified. This paper presents an algorithm for calculating the controller parameters of a grid-forming converter which guarantee a stable behavior for many different configurations of the grid.

A Multidimensional Data-Driven Sparse Identification Technique: The Sparse Proper Generalized Decomposition

Abstract: Sparse model identification by means of data is especially cumbersome if the sought dynamics live in a high dimensional space. This usually involves the need for large amount of data, unfeasible in such a high dimensional settings. This well-known phenomenon, coined as the curse of dimensionality, is here overcome by means of the use of separate representations. We present a technique based on the same principles of the Proper Generalized Decomposition that enables the identification of complex laws in the low-data limit. We provide examples on the performance of the technique in up to ten dimensions.

Sustainable-supplier selection for manufacturing services: a failure mode and effects analysis model based on interval-valued fuzzy group decision-making

Abstract: Inside supply chains’ exercises, evaluating suitable suppliers in light of the sustainability criteria, including economic, environmental, and social, can assist organizations to move toward sustainable development by considering their risks. Evaluating and choosing the sustainable-supplier for manufacturing services with lowest risks among candidates in the sustainable-supply chain management (S-SCM) is a vital issue for logistics managers, particularly by considering different sustainable criteria via three dimensions of the sustainability for strategic decisions. This paper introduces a new failure mode and effects analysis (FMEA) model based on multi-criteria decision-making by a group of supply chain-experts with interval-valued fuzzy (IVF) setting and asymmetric uncertainty information concurrently. In fact, the proposed model evaluates and ranks the suppliers according to their risks of economic, social, and environmental dimensions. Concepts of mean, variance, and skewness are introduced into the proposed FMEA model, and their mathematical relations are presented based on fuzzy possibilistic statistical concepts. Then, new definitions in the FMEA are presented for obtaining ideal solutions under uncertain conditions with possibilisic mean and possibilistic standard deviation, along with the possibilisic cube-root of skewness. Also, novel separation measures, max- and min-indices, and new fuzzy ranking index for risk scoring are presented to provide order of sustainable-supplier candidates under risky conditions. Finally, a real case study for manufacturing services is given and solved by the proposed FMEA model to demonstrate its capability in the S-SCM environment. The results of the proposed model illustrate that sustainable suppliers have been assessed and selected with the least amount of risks according to three dimensions of the sustainability.

The promises and pitfalls of Stochastic Gradient Langevin Dynamics

Abstract: Stochastic Gradient Langevin Dynamics (SGLD) has emerged as a key MCMC algorithm for Bayesian learning from large scale datasets. While SGLD with decreasing step sizes converges weakly to the posterior distribution, the algorithm is often used with a constant step size in practice and has demonstrated spectacular successes in machine learning tasks. The current practice is to set the step size inversely proportional to N where N is the number of training samples. As N becomes large, we show that the SGLD algorithm has an invariant probability measure which significantly departs from the target posterior and behaves like as Stochastic Gradient Descent (SGD). This difference is inherently due to the high variance of the stochastic gradients. Several strategies have been suggested to reduce this effect; among them, SGLD Fixed Point (SGLDFP) uses carefully designed control variates to reduce the variance of the stochastic gradients. We show that SGLDFP gives approximate samples from the posterior distribution, with an accuracy comparable to the Langevin Monte Carlo (LMC) algorithm for a computational cost sublinear in the number of data points. We provide a detailed analysis of the Wasserstein distances between LMC, SGLD, SGLDFP and SGD and explicit expressions of the means and covariance matrices of their invariant distributions. Our findings are supported by limited numerical experiments.

Natural α-Fe2O3 as an efficient catalyst for the p-nitrophenol reduction

Abstract: 4 nitrophenol (4-NP) is a recalcitrant water pollutant. It reduces to 4 aminophenol (4-AP), which is a less harmful compound. Typically, a noble metal catalyzes this reaction. However, its preparation is complex and expensive. Here, for the first time, we test hematite either from natural goethite calcination and compared with a synthetic hematite prepared by citrate or traditional precipitation method. We characterize the catalysts by thermo gravimetric analysis (TGA), X-ray fluorescence (XRF), X-ray diffraction (XRD) transmission electron microscopy (TEM), energy dispersive X-ray spectroscopy (EDX), particle size distribution (PSD), and nitrogen adsorption desorption at 77 K. 4-NP reacted in presence of NaBH4 as the reducing agent to 4-AP. Kinetic study and catalyst stability were performed. Particle morphology played an important role in the reaction. All samples exhibited high catalytic activity with first order kinetic constants of 0.024, 0.027, 0.037 s−1 for natural α-Fe2O3, hematite prepared by precipitation, and citrate method, respectively.

Probabilistic Prediction from Planning Perspective: Problem Formulation, Representation Simplification and Evaluation Metric

Abstract: Accurate probabilistic prediction for intention and motion of road users is a key prerequisite to achieve safe and high-quality decision-making and motion planning for autonomous driving. Typically, the performance of probabilistic predictions was only evaluated by learning metrics for approximation to the motion distribution in the dataset. However, as a module supporting decision and planning, probabilistic prediction should also be evaluated from decision and planning perspective. Moreover, the evaluation of probabilistic prediction highly relies on the problem formulation variation and motion representation simplification, which lacks a formal foundation in a comprehensive framework. To address such concerns, we provide a systematic and unified framework for the analysis of three under-explored aspects of probabilistic prediction: problem formulation, representation simplification and evaluation metric. More importantly, we address the omitted but crucial problems in the three aspects from decision and planning perspective. In addition to a review of learning metrics, metrics to be considered from planning perspective are highlighted, such as planning consequence of inaccurate and erroneous prediction, as well as violations of predicted motions to planning constraints. We address practical formulation variations of prediction problems, such as decision-maker view and blind view for viewpoint, as well as reactive prediction for interaction, so that decision and planning can be facilitated.

Visual Iconicity Across Sign Languages: Large-Scale Automated Video Analysis of Iconic Articulators and Locations.

Abstract: We use automatic processing of 120,000 sign videos in 31 different sign languages to show a cross-linguistic pattern for two types of iconic form–meaning relationships in the visual modality. First ...

Self-Improving System Integration - Status and Challenges after Five Years of SISSY

Abstract: The self-improving system integration (SISSY) initiative has emerged in recent years in response to a systems engineering trend towards the organisation of open, interconnected systems integrating a large set of heterogeneous and autonomous subsystems. Based on the idea to equip subsystems with capabilities to assess and maintain their own integration status within the overall system composition, a variety of concepts, techniques, and contributions have been proposed and fruitfully discussed at the particular events of the underlying workshop series. In this article, we summarise and categorise these research efforts and derive a roadmap towards full-scale SISSY systems.

Morphology and genomic hallmarks of breast tumours developed by ATM deleterious variant carriers.

Abstract: The ataxia telangiectasia mutated (ATM) gene is a moderate-risk breast cancer susceptibility gene; germline loss-of-function variants are found in up to 3% of hereditary breast and ovarian cancer (HBOC) families who undergo genetic testing. So far, no clear histopathological and molecular features of breast tumours occurring in ATM deleterious variant carriers have been described, but identification of an ATM-associated tumour signature may help in patient management. To characterise hallmarks of ATM-associated tumours, we performed systematic pathology review of tumours from 21 participants from ataxia-telangiectasia families and 18 participants from HBOC families, as well as copy number profiling on a subset of 23 tumours. Morphology of ATM-associated tumours was compared with that of 599 patients with no BRCA1 and BRCA2 mutations from a hospital-based series, as well as with data from The Cancer Genome Atlas. Absolute copy number and loss of heterozygosity (LOH) profiles were obtained from the OncoScan SNP array. In addition, we performed whole-genome sequencing on four tumours from ATM loss-of-function variant carriers with available frozen material. We found that ATM-associated tumours belong mostly to the luminal B subtype, are tetraploid and show LOH at the ATM locus at 11q22–23. Unlike tumours in which BRCA1 or BRCA2 is inactivated, tumours arising in ATM deleterious variant carriers are not associated with increased large-scale genomic instability as measured by the large-scale state transitions signature. Losses at 13q14.11-q14.3, 17p13.2-p12, 21p11.2-p11.1 and 22q11.23 were observed. Somatic alterations at these loci may therefore represent biomarkers for ATM testing and harbour driver mutations in potentially ‘druggable’ genes that would allow patients to be directed towards tailored therapeutic strategies. Although ATM is involved in the DNA damage response, ATM-associated tumours are distinct from BRCA1-associated tumours in terms of morphological characteristics and genomic alterations, and they are also distinguishable from sporadic breast tumours, thus opening up the possibility to identify ATM variant carriers outside the ataxia-telangiectasia disorder and direct them towards effective cancer risk management and therapeutic strategies.

SARA: Security Automotive Risk Analysis Method

Abstract: Connected and automated vehicles aim to improve the comfort and the safety of the driver and passengers. To this end, car manufacturers continually improve actual standardized methods to ensure their customers safety, privacy, and vehicles security. However, these methods do not support fully autonomous vehicles, linkability and confusion threats. To address such gaps, we propose a systematic threat analysis and risk assessment framework, SARA, which comprises an improved threat model, a new attack method/asset map, the involvement of the attacker in the attack tree, and a new driving system observation metric. Finally, we demonstrate its feasibility in assessing risk with two use cases: Vehicle Tracking and Comfortable Emergency Brake Failure.

Correlation between mechanical and microstructural properties of molybdenum nitride thin films deposited on silicon by reactive RF magnetron discharge

Abstract: Molybdenum nitride thin films were deposited on (100) silicon substrates by R.F. magnetron sputtering of a Mo target in a (Ar-N2) gas mixtures. The films were studied by Scanning Electron Microscopy (SEM), Energy Dispersive Spectroscopy (EDS) and X-ray diffraction. The nanomechanical properties have been determined by nanoindentation and Peak-Force Quantitative Nanomechanical Mapping (PF-QNM). The total internal stresses were determined by curvature measurements and the Stoney formula. As thin film composition influences the morphology, the stress state and the mechanical properties, modifications are expected in this study where the nitrogen content is tuned. The film exhibits a polycrystalline structure with preferred orientation along (111) plane. The increase of the nitrogen content in the coating (N/Mo =1.1) induces a broadening of the full width at half maximum (FWHM) of the (111) diffraction peak, which is attributed to the presence of smaller crystallites. The residual stress and mechanical properties variation were correlated to the structural transition from γ-Mo2N to hexagonal and cubic MoN. The results show a good agreement between the nanomechanical properties obtained by nanoindentation and PF-QNM.

A Lagrangian Relaxation-based Algorithm to Solve a Home Health Care Routing Problem

Abstract: Nowadays, a rapid growth in the rate of life expectancy can be seen especially in the developed countries. Accordingly, the population of elderlies has been increased. By another point of view, the number of hospitals, retirement homes along with medical staffs has not been grown with a same rate. Hence, Home Health Care (HHC) operations including a set of nurses and patients have been developed recently by both academia and health practitioners to consider elderlies’ preferences willing to receive their cares at their homes instead of hospitals or retirement homes. To alleviate the drawbacks of pervious works and make HHC more practical, this paper introduces not only a new mathematical formulation considering new suppositions in this research area but also a solution approach based on Lagrangian relaxation theory has been employed for the first time. The main strategy of used algorithm aims to fill the gap between the lower bound and upper bound of problem and finds a solution which has both optimality and feasibility properties. By generating a number of numerical examples, results show the performance of the proposed algorithm analyzed by different criteria as well as the efficiency of developed formulation through a set of sensitivity analyses.

High temperature durability of a bond-coatless plasma-sprayed thermal barrier coating system with laser textured Ni-based single crystal substrate

Abstract: Thermal barrier coating systems are usually build-up with bond coats to ensure a good adhesion of the ceramic top coat and to protect the substrate against oxidation and corrosion. Such system is often subjected to complex thermo-mechanical loading. Because of the very different damage processes encountered during service operations, a simplified system was investigated by removing the bond-coat. Recently adhesion bond strength was enhanced using laser surface texturing of the substrate in thermal spraying processes. Atmospheric plasma spray yttria-stabilized-zirconia thermal barrier coating system was deposited on the Ni-based AM1 single crystalline superalloy without bond coat. Adhesion bond strength was already increased compared to conventional processing method. Top coat durability was evaluated at high temperature and damage mechanisms were studied. Isothermal and cyclic oxidation tests showed durability of 1000 h and 400 cycles at 1100 °C. The oxidation mechanisms at the substrate/top coat interface changed due to fast solidification during the laser texturing process. Then, TBC system was studied under high temperature mechanical solicitation in tension creep. The textured interfaces were not damaged after 1% creep strain while top-coat/substrate interfacial cracking was observed for grit-blasted specimens. Moreover, no preferential crack development in the substrate was observed. Patterns provided an enhanced adhesion by changing the stress distribution near the interface.

Reduced-order modeling of soft robots

Abstract: We present a general strategy for the modeling and simulation-based control of soft robots. Although the presented methodology is completely general, we restrict ourselves to the analysis of a model robot made of hyperelastic materials and actuated by cables or tendons. To comply with the stringent real-time constraints imposed by control algorithms, a reduced-order modeling strategy is proposed that allows to minimize the amount of online CPU cost. Instead, an offline training procedure is proposed that allows to determine a sort of response surface that characterizes the response of the robot. Contrarily to existing strategies, the proposed methodology allows for a fully non-linear modeling of the soft material in a hyperelastic setting as well as a fully non-linear kinematic description of the movement without any restriction nor simplifying assumption. Examples of different configurations of the robot were analyzed that show the appeal of the method.

On the Link Between External Forcings and Slope Instabilities in the Piton de la Fournaise Summit Crater, Reunion Island

Abstract: We have analyzed the impact of different forcings, such as rain and seismicity, on slope instabilites on an active volcano. For this, we compiled a catalog of the locations and volumes of rockfalls in the Piton de la Fournaise crater using seismic records. We validated it by comparing the locations and volumes to those deduced from photogrammetric data. We analyzed 10,477 rockfalls, spanning the period 2014 to 2016. This period corresponds to the renewal of volcanic activity after a 41‐month rest. Our analysis reveals that renewed eruptive activity has unsettled the crater edges. External forcings such as rain and seismicity are shown to potentially increase the number and the volume of rockfalls, with a stronger impact on the volume. Preeruptive seismicity seems to be the main triggering factor for the largest volumes, with a delay of one to several days. Rain alone does not seem to trigger especially large rockfalls. We infer that repetitive vibrations from the many seismic events, combined with the action of rain, induce crack (or slip) growth in highly fractured (or granular) materials, leading to the collapse of large volumes. Regarding their spatial distribution before an eruption, the largest rockfalls seem to migrate toward the location of magma extrusion.

Impact of Perturbations on Nonlinear Frequency-Division Multiplexing

Abstract: Nonlinear frequency-division multiplexing (NFDM) is a communication scheme in which users’ signals are multiplexed in the nonlinear Fourier domain. The contributions of this paper are twofold. First, the achievable information rates (AIRs) of NFDM based on an integrable model of the optical fiber are summarized. For this ideal model, it is shown that the AIR of the NFDM is greater than the AIR of the wavelength-division multiplexing (WDM) for a given bandwidth and signal power, in a representative system with five users and one symbol per user. The improvement results from nonlinear signal multiplexing. Second, the impact of some of the main perturbations on NFDM are investigated, including the fiber loss, polarization effects, and the third-order dispersion. For a realistic nonideal model, it is shown that the WDM AIR with joint dual-polarization back-propagation and third-order dispersion compensation is approximately equal to the NFDM AIR with two independent single-polarization demodulations and without third-order dispersion compensation. Using a joint dual-polarization receiver and perturbations compensation is expected to increase the NFDM AIR.

The promises and pitfalls of Stochastic Gradient Langevin Dynamics.

Abstract: Stochastic Gradient Langevin Dynamics (SGLD) has emerged as a key MCMC algorithm for Bayesian learning from large scale datasets. While SGLD with decreasing step sizes converges weakly to the posterior distribution, the algorithm is often used with a constant step size in practice and has demonstrated successes in machine learning tasks. The current practice is to set the step size inversely proportional to $N$ where $N$ is the number of training samples. As $N$ becomes large, we show that the SGLD algorithm has an invariant probability measure which significantly departs from the target posterior and behaves like Stochastic Gradient Descent (SGD). This difference is inherently due to the high variance of the stochastic gradients. Several strategies have been suggested to reduce this effect; among them, SGLD Fixed Point (SGLDFP) uses carefully designed control variates to reduce the variance of the stochastic gradients. We show that SGLDFP gives approximate samples from the posterior distribution, with an accuracy comparable to the Langevin Monte Carlo (LMC) algorithm for a computational cost sublinear in the number of data points. We provide a detailed analysis of the Wasserstein distances between LMC, SGLD, SGLDFP and SGD and explicit expressions of the means and covariance matrices of their invariant distributions. Our findings are supported by limited numerical experiments.

Electron–Phonon Coupling in Luminescent Europium-Doped Hydride Perovskites Studied by Luminescence Spectroscopy, Inelastic Neutron Scattering, and First-Principles Calculations

Abstract: We present a case study on the vibrational coupling of lattice phonons to the electronic 4f7 (8S7/2)–4f65d1 (eg) transition of divalent europium in the hydrides and deuterides LiMH3 and LiMD3 (M = Sr and Ba). For low doping concentrations, these compounds show extraordinarily well-resolved vibronic fine structures at low temperatures. Besides luminescence emission spectroscopy of the europium-doped compounds, we carried out inelastic neutron scattering (INS) experiments of the europium-free compounds. The phonons coupling to the electronic transition are identified, and a good agreement between the vibronic and the INS data is found. The frequencies of the low-energy acoustic modes do not significantly change upon replacing hydride by deuteride, whereas a decrease by a factor of approximately 2 can be observed for the higher energy optic modes. Furthermore, we compare these experimental results to density functional calculations performed with the Vienna Ab initio Simulation Package. Knowledge of the phon...