In modern society, power consumption plays a crucial role, since it is capable of influencing multiple sectors including residential, commercial, and industrial domains. It covers the electrical energy amount used by various devices, appliances, machinery, and systems within a specific time frame. The accurate prediction of power consumption is imperative for effective energy management, resource allocation, infrastructure planning, and cost optimization. In this study, we focused on predicting power consumption within DAEWOO Steel CO. Ltd, located in South Korea. We acquired, preprocessed, and analyzed a dataset containing information about the daily operations of the industrial facility, with data being sampled at 15-minute intervals. By leveraging machine learning techniques, we employed six tree-based algorithms and three ensemble learners to forecast the target variable. Our comparative analysis examined the performance of these regressors across three forecasting horizons: 15 minutes, 1 hour, and 1 day. We discovered that the efficacy of the regressors is complex and is linked to the forecasting horizon. Notably, the stacking ensemble learner outperformed others for the 15-minute horizon, achieving impressive metrics of 98.5% for D2, 99.9% for R2, 0.81 for MSE, and 0.37 for MAE. For 1-hour ahead predictions, XGBoost emerged as the most accurate model, attaining metrics of 96.3% for D2, 99.7% for R2, 53.07 for MSE, and 3.5 for MAE. Finally, for 1-day ahead forecasting, the Extra Tree regressor surpassed its counterparts, achieving metrics of 93.1% for D2, 99.3% for R2, 13944 for MSE, and 74.46 for MAE. These findings underscore the importance of tailoring predictive models to specific forecasting horizons and highlight the efficacy of ensemble learning techniques in enhancing power consumption predictions across varying time frames.

With the ageing of the world population, Do-ItYourself approaches and Smart Personal Assistant are combined to provide customised systems that help people in their daily lives, including fostering their self-reliance and promoting ageing in place, even when these people have cognitive disabilities or perceptual impairments such as “Alzheimer” disease. Ambient Assisted Living systems which rely on this combination can also help caregivers reduce their workload by giving them much relevant information about the health status and the situation of the patients at home, and, furthermore, by giving them facilities to create home automation scenarios to assist the patients. A scenario is the arrangement of daily activities in order to achieve a particular goal. In this paper, we propose a new approach to creating assistance scenarios through voice command, involving two types of ontologies, a smart home ontology (OntoDomus) and a knowledge graph of an activity/action extracted from a scenario in the form of structured knowledge. In particular, we leverage ontology alignment to process voice requests in order to create an assistance scenario for the elderly.

This work delves into the significant impact of Machine Learning (ML) on the advancement and improvement of Energy Management Systems (EMS), focusing on the incorporation of renewable energy sources, smart grids, and the general enhancement of energy efficiency, reliability, and sustainability. The main aim of this work is to offer a detailed summary of Machine Learning technologies that can be used in modern energy systems. It explains how these technologies can improve certain tasks like load forecasting, energy optimization, predictive maintenance, fault detection and diagnosis, and incorporating renewable energy systems supported by relevant approaches and application areas. Moreover, the work examines the benefits and opportunities presented by machine learning in boosting efficiency, enhancing system stability and resilience, and contributing to environmental sustainability. In addition, it identifies challenges and outlines future research needed to facilitate the adoption of ML in energy systems. In conclusion, the study underlines the critical role of machine learning in the evolution of energy systems and underscores the importance of collaborative efforts to overcome existing challenges and fully leverage machine learning’s potential in the smart energy management systems domain.

Software technology has exponentially evolved leading to the development of intelligent applications using artificial intelligence models and techniques.Such development impacts all scientific and social fields: home automation, medicine, communication, etc.To make those new applications useful to a larger number of people, researchers are working on how to integrate artificial intelligence into real world while respecting the notion of calm technology.This paper fits in the context of the development of intelligent systems termed "wise systems" that aim at satisfying the calm technology requirement.Those systems are based on the concept of "Wise Object": a software entityobject, service, component, application, etc. -able to learn by itself how it is expected to behave and how it is used by a human or another software entity.During its learning process, a Wise Object constructs a graph that represents its behavior and the way it is used.A major weakness of Wise Objects is that the numerical information that they generate is mostly meaningless to humans.Therefore the objective of the work presented in this paper is to extend Wise Objects with semantic that enable them communicate with humans whose attention will consequently be less involved.In this paper, we address the issue of how to relate two different views using two state-based formalisms: State Transition Graph for views generated by the Wise Objects and Input Output Symbolic Transition System for conceptual views.Our proposal extends previous work done to extend the generated information with the conceptual knowledge using a matching algorithm founded on graph morphism.The first version of the algorithm has several limitations and constraints on the graphs that make it difficult to use in realistic cases.In this paper, we propose to generalize the algorithm and raise those restrictions.To illustrate the complete process, the construction of a sample graph matching on a home-automation system is considered.

In this paper, we propose a new method to combine displacement measurements from images with a low signal to noise ratio. The method takes advantage of the temporal redundancy of displacements that can be calculated from different image pairs to combine them in a single relative displacement time series robust to outliers. The method has only two parameters which determines the smoothness of the result. The proposed algorithm has been tested on displacements calculated from ground based stereo images of the Laurichard rock glacier. On the test data, our method outperforms the traditional inversion method by providing relative displacement time series with negligible noise and no outlier.

Data management has become crucial. More and more distributed data management solutions arise, e.g. Casandra or Cosmos DB. Some of them propose multiple consistency protocols. Thus, for each piece of data, the developer can choose a consistency protocol adapted to his needs. In this paper we explain why taking the consistency protocol into account is important while replicating data; and we propose CAnDoR, an approach that dynamically adapts the replication according to the data usage (read/write frequencies and locations) and the consistency protocol used to manage the piece of data. Our simulations show that using CAnDoR to place and move data copies can improve the global average access latency by up to 40%.

International audience

At the end of the 20th century, the development of spatial geodetic techniques (optical & SAR imagery, GPS) has allowed for drastic improvement of the spatial coverage and the resolution of the displacement measurements. The arrival of these techniques has caused an effective revolution by significantly improving our ability to measure the ground movement, as well as their temporal evolutions with great precision over large areas. Spectacular results have been obtained in numerous applications with displacement of various characteristics (in terms of magnitude, duration, spatial distribution): the study of subsidence in urban areas, of the co-seismic, inter-seismic and post-seismic motions, of glacier flows, of volcanic deformation, etc. Nowadays, the displacement maps obtained by remote sensing techniques cover almost the whole world, with a precision within millimetres per year. Therefore, they are considered as the predominant sources for studies of the terrestrial deformation, from which geophysical models of the deformation have been retrieved to further understand the deformation source in depth. To this end, good knowledge of the reliability of the remote sensing data, as well as of the physical models accordingly obtained is crucial for all the researches and applications that use these sources of information. A perspective of significant improvement in the accuracy of the displacement measurement appears with the growing availability of remote sensing data. Methodological development in fusion of displacement measurements and of the integration of a physical model based on the supercomputer facilities seems necessary to reduce the uncertainty and to improve the accuracy of the displacement measurement. In this context, this paper addresses the current status, challenges and perspectives of the remotely sensed displacement measurement fusion.

In this paper, an attempt of Alpine glacier flow modeling is performed based on a series of high resolution TerraSAR-X SAR images and a Digital Elevation Model. First, a glacier flow model is established according to the fluid mechanics theory in a simplified framework. Second, the displacement field over the glacier obtained from the sub-pixel image correlation of a series of TerraSAR-X SAR images is used to refine the model obtained previously. The comparison between the data observation and the model prediction allows for the validation of the established model. According to the obtained results, despite the simplifications made in the modeling, the established glacier flow model can provide general satisfactory results. Further investigation and improvement of this glacier flow model seem promising.

This paper presents the first results obtained with satellite image time series (SITS) acquired by Cosmo-SkyMed (CSK) over the Chamonix Mont-Blanc test-site. A CSK SITS made of 39 images is merged with a TerraSAR-X SITS made of 26 images by using the orbital information and co-registration tools developed in the EFIDIR project. The results are illustrated by the computation of speckle-free images by temporal averaging, by the generation and comparison of topographic interferograms and by the measure of glacier displacement fields by amplitude correlation.

Domotic systems aim to offer functionalities like energy management, security, conveniences, and much more. Many domotic networks exist to provide a subset of these applications, but these networks are not necessarily compatible due to different communication mediums or protocols. Literature presents different studies that introduce high-level systems that solve the lack of incompatibility, but it does not explore how to create a network behavior. This paper concentrates on studying how to model the unit behaviors of home devices and a global behavior of a network of these devices: the automated living area. The global behavior is set-up with rules and constraints. The behaviors are modeled with an extended automaton input-output symbolic transition systems. To finish this paper, a use case shows the interest of building global behavior with unit behaviors.

This paper investigate a mutual exclusion algorithm on distributed systems. We introduce a new algorithm based on the Naimi-Trehel algorithm, taking advantage of the distributed approach of Naimi-Trehel while allowing to request partial locks. Such ranged locks offer a semantic close to POSIX file locking, where threads lock some parts of the shared file. We evaluate our algorithm by comparing its performance with to the original Naimi-Trehel algorithm and to a centralized mutual exclusion algorithm. The considered performance metric is the average time to obtain a lock.

In this paper, first order diffusion load balancing algorithms for totally dynamic networks are investigated. Totally dynamic networks are networks in which the topology may change dynamically. Some edges or nodes can appear, disappear or move during the time. In our previous works on dynamic networks, the dynamism was limited to the edges. The main result of this study consists in proving that the load balancing algorithms reduce the unbalance on arbitrary dynamic networks. Notice that the hypotheses of our result are realistic and that for example the network does not have to be maintained connected. To study the behavior of these algorithms, we compare the load evolution by several simulations.

We introduce a theoretical algorithm and its practical version to perform a decentralized detection of the global convergence of parallel asynchronous iterative algorithms. We prove that, even if the algorithm is completely decentralized, the detection of global convergence is achieved on one processor under the classical conditions. The proposed algorithm is very useful in the context of grid computing in which the processors are distributed and in which detecting the convergence on a master processor may be penalizing or even impossible as in peer to peer computation frameworks. Finally, the efficiency of the practical algorithm is illustrated in a typical experiment.