Search

In this paper, three challenges often encountered when upper limb rehabilitation robots are integrated with impaired people are addressed. Firstly, estimation of Desired Intended Motion (DIM) of the robot’s wearer is achieved. Secondly, robust adaptive impedance control based on the Modified Function Approximation Technique (MFAT) is designed. Lastly, a new Integral Nonsingular Terminal Sliding Mode Control (INTSMC) is suggested. In particular, the integration of INTSMC enriches the system by ensuring continuous performance tracking of system’s trajectories, high robustness, fast transient response, finite-time convergence, and chattering reduction. Besides, the MFAT strategy approximates the dynamic model without collecting any prior knowledge of the lower and upper bounds of the dynamic model’s individual uncertainties. Furthermore, leveraging Radial Basis Function Neural Network (RBFNN) to link estimated DIM to the adaptive impedance control allows the upper limb robot to easily track the target impedance model. Finally, in efforts to validate the scheme in real-time, controlled experimental cases are conducted using the exoskeleton robot.

In this paper, a walking algorithm of a biped robot with unknown dynamics has been addressed by proposing a new adaptive control based on the Modified Function Approximation Technique (MFAT) augmented with backstepping control, using state feedback and output feedback. The classical ZMP, the preview controller on the 3- Dimensional Linear Inverted Pendulum Model (3D-LIPM), is a stable walking algorithm but does not allow high accuracy, high speed, and robustness due to the ground environment, and also due to the dynamics uncertainties caused by the complex structure of these kinds of robots. To improve the accuracy of the walking tracking performance of the biped robot and its robustness against dynamics uncertainty and external perturbations, a combination of backstepping control and MFAT strategy have been intended. Recursive Lyapunov function is developed to ensure the closed-loop system stability. Experiment simulation and comparative study were carried out with biped robot to verify the performance and the robustness of the developed control strategy.

This paper presents the design and validation of a new adaptive variable gain reaching law, integrated with sliding mode control (SMC), to control perturbed and unperturbed nonlinear systems. The novelty behind this law stems from its capability to overcome the main limitations involved with SMC. In contrast to existing reaching laws, system’s performance can be substantially enhanced via this law, with significant reduction in the chattering phenomenon, along ensuring rapid convergence time of system’s trajectories towards equilibrium. The designed law not only integrates the features of both the exponential reaching law (ERL) and the power rate reaching law (PRL), but also overcomes their limitations. Simulation and comparison studies against ERL and PRL were carried out to validate the effectiveness and advantages of the proposed reaching law scheme (Proposed-RL). Furthermore, controlled experimental investigations were conducted using an exoskeleton robot (ETS-MARSE) to validate the scheme in real-time.

Clustering is a crucial step in scientific data analysis and engineering systems. Thus, an efficient cluster analysis method often remains a key challenge. In this paper, we introduce a general purpose exemplar-based clustering method called (MEGA), which performs a novel message-passing strategy based on variational expectation–maximization and generalized arc-consistency techniques. Unlike message passing clustering methods, MEGA formulates the message-passing schema as E- and M-steps of variational expectation–maximization based on a reparameterized factor graph. It also exploits an adaptive variant of generalized arc consistency technique to perform a variational mean-field approximation in E-step to minimize a Kullback–Leibler divergence on the model evidence. Dissimilar to density-based clustering methods, MEGA has no sensitivity to initial parameters. In contrast to partition-based clustering methods, MEGA does not require pre-specifying the number of clusters. We focus on the binary-variable factor graph to model the clustering problem but MEGA is applicable to other graphical models in general. Our experiments on real-world problems demonstrate the efficiency of MEGA over existing prominent clustering algorithms such as Affinity propagation, Agglomerative, DBSCAN, K-means, and EM.

This paper presents the design, and validation of a new adaptive control system based on quasi-time delay estimation (QTDE) augmented with new integral second-order terminal sliding mode control (ISOTSMC) for a manipulator robot with unknown dynamic uncertainty and disturbances. Contrary to the conventional TDE, the proposed Q-TDE becomes sufficient to invoke a fixed artificial time delay and utilize the past data only of the control input to approximate the unknown system’s dynamic uncertainties. The incorporating of new adaptive reaching law with ISOTSMC augmented with Q-TDE policy ensures the continuous performance tracking of the robot manipulator’s trajectories using output feedback. This combination may achieve high performance with a significant chattering reducing procedure. By utilizing the Lyapunov function theory, it can be demonstrated that the robot system is stable and all signals in closed-loop are converging in finite time. Consequently, Simulation and comparative studies with two degrees of freedom robot manipulator were carried out to validate the effectiveness of the designed control scheme.

A wide range of important problems in machine learning, expert system, social network analysis, bioinformatics and information theory can be formulated as a maximum a-posteriori (MAP) inference problem on statistical relational models. While off-the-shelf inference algorithms that are based on local search and message-passing may provide adequate solutions in some situations, they frequently give poor results when faced with models that possess high-density networks. Unfortunately, these situations always occur in models of real-world applications. As such, accurate and scalable maximum a-posteriori (MAP) inference on such models often remains a key challenge. In this paper, we first introduce a novel family of extended factor graphs that are parameterized by a smoothing parameter χ ∈ [0,1]. Applying belief propagation (BP) message-passing to this family formulates a new family of W eighted S urvey P ropagation algorithms (WSP-χ) applicable to relational domains. Unlike off-the-shelf inference algorithms, WSP-χ detects the “backbone” ground atoms in a solution cluster that involve potentially optimal MAP solutions: the cluster backbone atoms are not only portions of the optimal solutions, but they also can be exploited for scaling MAP inference by iteratively fixing them to reduce the complex parts until the network is simplified into one that can be solved accurately using any conventional MAP inference method. We also propose a lazy variant of this WSP-χ family of algorithms. Our experiments on several real-world problems show the efficiency of WSP-χ and its lazy variants over existing prominent MAP inference solvers such as MaxWalkSAT, RockIt, IPP, SP-Y and WCSP.

Discovering influential nodes (or actors) in the network is often the key task of mining, analyzing, and understanding real-life networks. Centrality measures are commonly used to detect important nodes that control the information propagation in the network. While off-the-shelf centrality indices may provide effective node identification in several situations, they frequently produce inadequate results when confronted with massive networks, in the presence of complex local structures or the lack of certain characteristics. In this paper, we introduce Cross-face, a new scalable centrality measurement for the identification of key nodes in such networks. Inspired by the Formal Concept Analysis (FCA) framework, the conceptual idea of “Cross-face” is to leverage the faces of concepts to identify nodes that are located in “face bridges” and have an influential “cross clique” connectivity. Thus, it concurrently measures how the node influences its neighbour nodes through its cross cliques while linking the densely connected substructures of the network via its presence in bridges. Unlike traditional centrality measures, the cross-face of nodes can be computed using only a set of symmetrical concepts, which is often quite small compared to the set of nodes or edges in the network. Our experiments on several real-world networks show the efficiency of Cross-face over existing prominent centrality indices such as betweenness, closeness, eigenvector, and k-shell among others.

Rehabilitation robots have shown a high potential for improving the patients’ mobility, improving their functional movements and assisting in daily activities. However, this technology is still an emerging field and suffers from several challenges like compliance control and dynamic uncertain caused by the human–robot collaboration. The main challenge addressed in this paper is to ensure that the exoskeleton robot provides a suitable compliance control that allows it to cooperate perfectly with humans even if the dynamic model of the exoskeleton robot is uncertain. To achieve that, an adaptive tracking controller based on Modified Function Approximation Technique (FAT) is proposed to approximate the dynamic model of the exoskeleton robot. Unlike a conventional FAT, the required use of basis functions in estimations law of dynamic model and the acceleration feedback is eliminated in the proposed modified FAT. Additionally, the desired trajectory is designed based on the designer’s prediction of the motion intention of the human, using the Damped Least Square method in order to reduce the error between the actual position of the robot and the motion intention of the human which help the subject move the exoskeleton arm easily in active rehabilitation tasks. The stability analysis is formulated and demonstrated based on Lyapunov function. An experimental physiotherapy session and comparison study with a healthy subject was performed to test the effectiveness and robustness of the proposed adaptive control.

This paper deals with the optimal energy management system (EMS) in wind-based stand-alone microgrid. A new day-ahead integrated control based on dynamic programming (DP) is proposed to optimize the power flow and enhance the battery's lifetime. Defined constraints help to set system limits for safe operation. Finally, the economic model allow us to define the parameters of the system to be optimized. Simulation results using MATLAB/SIMULINK show the effectiveness of the proposed EMS.

Discovering meaningful conceptual structures is a substantial task in data mining and knowledge discovery applications. While off-the-shelf interestingness indices defined in Formal Concept Analysis may provide an effective relevance evaluation in several situations, they frequently give inadequate results when faced with massive formal contexts (and concept lattices), and in the presence of irrelevant concepts. In this paper, we introduce the Conceptual Relevance (\(\mathcal {CR}\)) score, a new scalable interestingness measure for the identification of relevant concepts. From a conceptual perspective, minimal generators provide key information about their associated concept intent. Furthermore, the relevant attributes of a concept are those that maintain the satisfaction of its closure condition. Thus, \(\mathcal {CR}\) exploits the fact that minimal generators and relevant attributes can be efficiently used to assess concept relevance. As such, the \(\mathcal {CR}\) index quantifies both the amount of conceptually relevant attributes and the number of the minimal generators per concept intent. Our experiments on synthetic and real-world datasets show the efficiency of this measure over the well-known stability index.

Identifying important actors (or nodes) in a two-mode network is a crucial challenge in mining, analyzing, and interpreting real-world networks. While traditional bipartite centrality indices are often used to recognize key nodes that influence the network information flow, inaccurate results are frequently obtained in intricate situations such as massive networks with complex local structures or a lack of complete knowledge about the network topology and certain properties. In this paper, we introduce Bi-face (BF), a new bipartite centrality measurement for identifying important nodes in two-mode networks. Using the powerful mathematical formalism of Formal Concept Analysis, the BF measure exploits the faces of concept intents to detect nodes that have influential bicliques connectivity and are not located in irrelevant bridges. Unlike off-the shelf centrality indices, it quantifies how a node has a cohesive substructure influence on its neighbour nodes via bicliques while not being in network core-peripheral ones through its absence from non-influential bridges. In terms of identifying accurate node centrality, our experiments on a variety of real-world and synthetic networks show that BF outperforms several state-of-the art bipartite centrality measures, producing the most accurate Kendall coefficient. It provides unique node identification based on network topology. The findings also demonstrate that the presence of terminal nodes, influential bridges, and overlapping key bicliques impacts both the performance and behaviour of BF as well as its relationship with other traditional centrality measures. On the datasets tested, the computation of BF is at least twenty-three times faster than betweenness, eleven times faster than percolation, nine times faster than eigenvector, and ten times faster than closeness.

6th IEEE International Energy Conference (IEEE ENERGYCON) - Energy Transition for Developing Smart Sustainable Cities, IEEE, ELECTR NETWORK, SEP 28-OCT 01, 2020; International audience; The intermittent nature of solar energy creates a significant challenge for the optimization and planning of future smart grids. In order to reduce intermittency, it is very important to accurately predict Photovoltaic (PV) power generation. This work proposes a new prediction method based on the Recurrent Neural Network (RNN) for accurately predicting the yield of photovoltaic power generation systems. Our study used a Longe Short-Term Memory (LSTM) architecture. The LSTM approach can store information over time, which is valuable for time series prediction. The proposed prediction method is evaluated using real PV energy in Lille, France. Firstly, all solar time series data are divided into three main parts: 70% of the data are used to train the neural network, 20% of the data are used for verification and the other data are used for testing. The proposed prediction method has a good prediction quality in very short term (one-hour), which proves the reliability and cost-effectiveness of this method.

7th International Conference on Control, Decision and Information Technologies (CoDIT), Prague, CZECH REPUBLIC, JUN 29-JUL 02, 2020; International audience; Fluctuating photovoltaic (PV) power creates a challenge for grid energy management as energy penetration increases rapidly. As a result, the forecasting of PV energy has become essential to ensure grid stability and promote optimal use of energy resources and more economical distribution. The objective of this work is to forecast PV power for a 10 min horizon, based on a set of data. This approach is used to ensure efficient operation of the fluctuating output of a PV station charging a station for electic vehicules. Simulation results show the reliability and efficiency of the technique.

The originality of this paper consists of analyzing and comparing the installation of and Photovoltaic Dish Stirling (DS) systems in buildings. The main goal is to maximize energy production using only the available building rooftop by introducing different solar technologies. For the case study, a typical building with four floors located in Lebanon is considered. The main goal is to propose a self-sufficient energy building that produces its own energy without the need to be connected to the grid or supplied by other external electricity sources. Three case studies are considered in which different solar technologies are installed on the rooftop: (i) PV system, (ii) Dish Stirling System, (iii) combination of (i), and (ii) to maximize the energy generation and the utilization factor of the rooftop. Results show that the PV system of 9. 3kW can only supply 12% of the total energy demand of the building during a day, while 68% is attained by using one Dish Stirling of a capacity of 25kWp. In the third case, the combination of both systems yields about 73% of the total energy demand of the building.

Distributed Generation (DG) systems, such as renewable energy sources and fuel cells are widely used for standalone microgrids to complement fossil fuel generators. However, these systems still suffer from various problems such as dynamics uncertainties, external disturbances, and imprecision of the system parameters. In this paper, we present a new Integral super-twisting Terminal Sliding Mode Control incorporating Time Delay Estimation (STA-TDE) applied to Three-Phase Inverter for Standalone Distributed Generation System with dynamics uncertainties and unknown bounded disturbances. Unlike conventional adaptive approaches, the proposed Time Delay Estimation uses delayed one step only of the control input and outputs of the system to approximate the uncertain dynamics. A finite time of both selected sliding surface and estimation error simultaneous is achieved using an appropriate Lyapunov function. Numerical simulation results found to confirm the effectiveness of the proposed control.

In this paper, a new sliding mode control (SMC) is applied to a physical nonlinear system. The novelty of this approach is related to the proposed reaching law by overcoming the main limitations of SMC. Unlike existing reaching laws, the suggested one can achieve high performance with significant reducing of a chattering problem and has a very fast convergence time of the system trajectories into the origin. This law benefits from the advantages and overcomes the limitations of both the exponential reaching law (ERL) and the conventional sliding mode control (SMC). Simulation results and comparison study with ERL and SMC are presented and applied on two degrees of freedom robot in order to show the advantage of the proposed adaptive reaching law. Experiments results are performed with electric cylinder (DC Motor) to confirm this proposition in real-time implementation.

Many important real-world applications of machine learning, statistical physics, constraint programming and information theory can be formulated using graphical models that involve determinism and cycles. Accurate and efficient inference and training of such graphical models remains a key challenge. Markov logic networks (MLNs) have recently emerged as a popular framework for expressing a number of problems which exhibit these properties. While loopy belief propagation (LBP) can be an effective solution in some cases; unfortunately, when both determinism and cycles are present, LBP frequently fails to converge or converges to inaccurate results. As such, sampling based algorithms have been found to be more effective and are more popular for general inference tasks in MLNs. In this paper, we introduce Generalized arc-consistency Expectation Maximization Message-Passing (GEM-MP), a novel message-passing approach to inference in an extended factor graph that combines constraint programming techniques with variational methods. We focus our experiments on Markov logic and Ising models but the method is applicable to graphical models in general. In contrast to LBP, GEM-MP formulates the message-passing structure as steps of variational expectation maximization. Moreover, in the algorithm we leverage the local structures in the factor graph by using generalized arc consistency when performing a variational mean-field approximation. Thus each such update increases a lower bound on the model evidence. Our experiments on Ising grids, entity resolution and link prediction problems demonstrate the accuracy and convergence of GEM-MP over existing state-of-the-art inference algorithms such as MC-SAT, LBP, and Gibbs sampling, as well as convergent message passing algorithms such as the concave---convex procedure, residual BP, and the L2-convex method.

One key challenge in Formal Concept Analysis is the scalable and accurate computation of stability index as a means to identify relevant formal concepts. Unfortunately, most exact methods for computing stability have an algorithmic complexity that could be exponential w.r.t. the context size. While randomized approximate algorithms, such as Monte Carlo Sampling (MCS), can be good solutions in some situations, they frequently lead to the slow convergence problem with an inaccurate estimation of stability. In this paper, we introduce a new approximation method to estimate the stability using the low-discrepancy sampling (LDS) approach. To improve the convergence rate, LDS uses quasi-random sequence to distribute the sample points evenly across the power set of the concept intent (or extent). This helps avoid the clumping of samples and let all the areas of the sample space be duly represented. Our experiments on several formal contexts show that LDS can achieve faster convergence rate and better accuracy than MCS.

We address highly dynamic distributed systems modelled by time-varying graphs (TVGs). We are interested in proof of impossibility results that often use informal arguments about convergence. First, we provide a topological distance metric over sets of TVGs to correctly define the convergence of TVG sequences in such sets. Next, we provide a general framework that formally proves the convergence of the sequence of executions of any deterministic algorithm over TVGs of any convergent sequence of TVGs. Finally, we illustrate the relevance of the above result by proving that no deterministic algorithm exists to compute the underlying graph of any connected-over-time TVG, i.e. Any TVG of the weakest class of long-lived TVGs.

We address the problem of computing a Minimal Dominating Set in highly dynamic distributed systems. We assume weak connectivity, i.e., the network may be disconnected at each time instant and topological changes are unpredictable. We make only weak assumptions on the communication: every process is infinitely often able to communicate with other processes (not necessarily directly). Our contribution is threefold. First, we propose a new definition of minimal dominating set suitable for the context of time-varying graphs that seems more relevant than existing ones. Next, we provide a necessary and sufficient topological condition for the existence of a deterministic algorithm for minimal dominating set construction in our settings. Finally, we propose a new measure of time complexity in time-varying graph in order to to allow fair comparison between algorithms. Indeed, this measure takes account of communication delays attributable to dynamicity of the graph and not to the algorithms., arXiv admin note: text overlap with arXiv:1412.6007

International audience; We consider highly dynamic distributed systems modelled by time-varying graphs (TVGs). We first address proof of impossibility results that often use informal arguments about convergence. We provide a general framework that formally proves the convergence of the sequence of executions of any deterministic algorithm over TVGs of any convergent sequence of TVGs. Next, we focus of the weakest class of long-lived TVGs, i.e., the class of TVGs where any node can communicate any other node infinitely often. We illustrate the relevance of our result by showing that no deterministic algorithm is able to compute various distributed covering structure on any TVG of this class. Namely, our impossibility results focus on the eventual footprint, the minimal dominating set and the maximal matching problems.