Solar energy is a vast and clean resource that can be harnessed with great benefit for humankind. It is still currently difficult, however, to convert it into electricity in an efficient and cost-effective way. One of the ways to produce energy is the use of various focusing technologies that concentrate the direct normal irradiance (DNI) to produce power through highly-efficient modules or conventional turbines. Concentrating technologies have great potential over arid areas, such as Northern Africa. A serious issue is that DNI can vary rapidly under broken-cloud conditions, which complicate its forecasts [1]. In comparison, the global horizontal irradiance (GHI) is much less sensitive to cloudiness. As an alternative to the direct DNI forecasting avenue, a possibility exists to derive the future DNI indirectly by forecasting GHI first, and then use a conventional separation model to derive DNI. In this context, the present study compares four of the most well-known separation models of the literature and evaluates their performance at Tamanrasset, Algeria, when used in combination with a new deep learning machine methodology introduced here to forecast GHI time series for short-term horizons (15-min). The proposed forecast system is composed of two separate blocs. The first bloc seeks to forecast the future value of GHI based on historical time series using the Long Short-Term Memory (LSTM) technique with two different search algorithms. In the second bloc, an appropriate separation (also referred to as “diffuse fraction” or “splitting”) model is implemented to extract the direct component of GHI. LSTMs constitute a category of recurrent neural network (RNN) structure that exhibits an excellent learning and predicting ability for data with time-series sequences [2]. The present study uses and evaluates the performance of two novel and competitive strategies, which both aim at providing accurate short-term GHI forecasts: Unidirectional LSTM (UniLSTM) and Bidirectional LSTM (BiLSTM). In the former case, the signal propagates backward or forward in time, whereas in the latter case the learning algorithm is fed with the GHI data once from beginning to the end and once from end to beginning. One goal of this study is to evaluate the overall advantages and performance of each strategy. Hence, this study aims to validate this new approach of obtaining 15- min DNI forecasts indirectly, using the most appropriate separation model. An important step here is to determine which model is suitable for the arid climate of Tamanrasset, a high-elevation site in southern Algeria where dust storms are frequent. Accordingly, four representative models have been selected here, based on their validation results [3] and popularity: 1) Erbs model [4]; 2) Maxwell’s DISC model [5]; 3) Perez’s DIRINT model [6]; and 4) Engerer2 model [7]. In this contribution, 1-min direct, diffuse and global solar irradiance measurements from the BSRN station of Tamanrasset are first quality-controlled with usual procedures [3, 8] and combined into 15-min sequences over the period 2013–2017. The four separation models are operated with the 15-min GHI forecasts obtained with each LSTM model, then compared to the 15-min measured DNI sequences. Table 1 shows the results obtained by the two forecasting strategies, for the experimental dataset.

In the last years, predictive maintenance has gained a central position in condition-based maintenance tasks planning. Machine learning approaches have been very successful in simplifying the construction of prognostic models for health assessment based on available historical labeled data issued from similar systems or specific physical models. However, if the collected samples suffer from lack of labels (small labeled dataset or not enough samples), the process of generalization of the learning model on the dataset as well as on the newly arrived samples (application) can be very difficult. In an attempt to overcome such drawbacks, a new deep supervised learning approach is introduced in this paper. The proposed approach aims at extracting and learning important patterns even from a small amount of data in order to produce more general health estimator. The algorithm is trained online based on local receptive field theories of extreme learning machines using data issued from a propulsion system simulator. Compared to extreme learning machine variants, the new algorithm shows a higher level of accuracy in terms of approximation and generalization under several training paradigms.

Optical burst switching (OBS) has become one of the best and widely used optical networking techniques. It offers more efficient bandwidth usage than optical packet switching (OPS) and optical circuit switching (OCS).However, it undergoes more attacks than other techniques and the Classical security approach cannot solve its security problem. Therefore, a new security approach based on machine learning and cloud computing is proposed in this article. We used the Google Colab platform to apply Support Vector Machine (SVM) and Extreme Learning Machine (ELM)to Burst Header Packet (BHP) flooding attack on Optical Burst Switching (OBS) Network Data Set.

Industry 4.0 is a tsunami that will invade the whole world. The real challenge of the future factories requires a high degree of reliability both in machinery and equipment. Thereupon, shifting the rudder towards new trends is an inevitable obligation in this fourth industrial revolution where the maintenance system has radically changed to a new one called predictive maintenance 4.0 (PdM 4.0). This latter is used to avoid predicted problems of machines and increase their lifespan taking into account that if machines have not any predicted problem, they will never be checked. However, in order to get successful prediction of any kind of problems, minimizing energy and resources consumption along with saving costs, this PdM 4.0 needs many new emerging technologies such as the internet of things infrastructure, collection and distribution of data from different smart sensors, analyzing/interpreting a huge amount of data using machine/deep learning…etc. This paper is devoted to present the industry 4.0 and its specific technologies used to ameliorate the existing predictive maintenance strategy. An example is given via a web platform to get a clear idea of how PdM 4.0 is applied in smart factories.

In this paper, our objective is dedicated to the detection of a deterioration in the estimated operating time by giving preventive action before a failure, and the classification of breakdowns after failure by giving the action of the diagnosis and / or maintenance. For this reason, we propose a new Neuro-fuzzy assistance prognosis system based on pattern recognition called "NFPROG" (Neuro Fuzzy Prognosis). NFPROG is an interactive simulation software, developed within the Laboratory of Automation and Production (LAP) -University of Batna, Algeria. It is a four-layer fuzzy preceptor whose architecture is based on Elman neural networks. This system is applied to the cement manufacturing process (cooking process) to the cement manufacturing company of Ain-Touta-Batna, Algeria. And since this company has an installation and configuration S7-400 of Siemens PLC PCS7was chosen as a programming language platform for our system.

 Inventory management in distribution networks remains a challenging task due to the demand nature and the limited storage capacity. In this work, we study a three-level, a multi-product and a multi-period distribution network consisting of a central warehouse, three distribution centres and six wholesalers. Each of them faces a random demand. In order to optimise the inventory management in the distribution network, we first propose to make a horizontal cooperation between actors of the same level in the form of product exchange; then we propose a second approach based on vertical-horizontal cooperation. Both approaches are modelled as a MIP model and solved using the CPLEX solver. The objective of this study is to analyse the performance in terms of costs, quantities in stock and customer satisfaction.

Au cours des dernières années, les entreprises sont émergées dans un environnement concurrentiel avancé. Afin de répondre aux exigences de réduction des coûts, la demande des clients, les délais imposés, la qualité et l'amélioration de la variété, les entreprises doivent améliorer leur performance pour rester compétitives, survivre et se développer. Pour atteindre cet objectif, plusieurs modèles sont utilisés comme TQM (Total Quality Management), Kaizen, JAT (Just A Time), ERP (Enterprise Resource Planning), BPR (Business Process Reengineering) et Six Sigma, etc. Dans ce travail, nous nous sommes focalisés sur des modèles efficaces (aspirés de l’approche Six Sigma,) utilisés principalement pour justifier la compétitivité d'une entreprise. On s’intéresse particulièrement aux modèles DMAIC (Définir, Mesurer, Analyser, Innover et Contrôler, le modèle DPMO pondéré (défauts par million d'Opportunité) et au modèle ROF (Rendement optimal des flux. Ces modèles sont appliqués pour mesurer les niveaux de processus et d'évaluer la compétitivité de l'entreprise. Les résultats de ces modèles sont appliqués dans deux systèmes industriels de fabrication du ciment et des bouteilles à gaz.

Renewable energies offer new solutions to an ever-increasing energy demand. Wind energy is one of the main sources of electricity production, which uses winds to be converted to electrical energy with lower cost and environment saving. The major failures of a wind turbine occur in the bearings of high-speed shafts. This paper proposes the use of optimized machine learning to predict the Remaining Useful Life (RUL) of bearing based on vibration data and features extraction. Significant features are extracted from filtered band-pass of the squared raw signal where the health indicators are automatically selected using relief technique. Optimized Adaptive Neuro Fuzzy Inference System (ANFIS) by Partical Swarm Optimization (PSO) is used to model the non linear degradation of the extracted indicators. The proposed approach is applied on experimental setup of wind turbine where the results show its effectiveness for RUL estimation.

In this paper we address the integration of two-level supply chain with multiple items, production facility and retailers’ demand over a considered discrete time horizon. This two-level production distribution system features capacitated production facility supplying several retailers located in the same region. If production does take place, this process incurs a fixed setup cost as well as unit production costs. In addition, deliveries are made from the plant to the retailers by a limited number of capacitated vehicles and routing costs are incurred. This work aims at implementing a solution to minimize the sum of the costs at the production facility and the retailers. The methodology adopted to tackle this issue is based on a hybrid heuristic, greedy and genetic algorithms that uses strong formulation to provide a good solution of a guaranteed quality that are as good or better than those provided by the MIP optimizer with a considerably larger run time. The results demonstrate that the proposed heuristics are effective and performs impressively in terms of computational efficiency and solution quality.

In this paper, we address the integration of a two-level supply chain with multiple items. This two-level production-distribution system features a capacitated production facility supplying several retailers located in the same region. If production does occur, this process incurs a fixed setup cost and unit production costs. Besides, deliveries are made from the plant to the retailers by a limited number of capacitated vehicles, routing costs incurred. This work aims to implement a minimization solution that reduces the total costs in both the production facility and retailers. The methodology adopted based on a hybrid heuristic, greedy and genetic algorithm uses strong formulation to provide a suitable solution of a guaranteed quality that is as good or better than those provided by the MIP optimizer. The results demonstrate that the proposed heuristics are effective and performs impressively in terms of computational efficiency and solution quality.

L’objectif de ce travail de thèse et de développer des nouvelles approches permettant aux petites et moyennes entreprises d’améliorer les performances de leur processus de fabrication. Nous avons développé trois approches aspirées du Lean Six Sigma (LSS) pour l’amélioration de la production dans un contexte conventionnel et classique d’une part et d’autre part dans un contexte de production durable. Dans la première approche nous avons proposé une approche Lean Six Sigma conventionnelle pour évaluer et suivre la compétitivité d’une PME en fonction des résultats obtenus par la méthode VSM. Dans la deuxième approche, nous avons proposé une nouvelle extension de l’approche LSS vers le contexte de la production durable en incorporant des algorithmes multicritères quantitatives. Cette approche nous a permis de surmonter quelques barrières au niveau du processus de l’application du LSS. Dans La troisième approche nous avons présenté une amélioration de l’approche LSS qui vise à montrer l’effet positif des algorithmes multicritères qualitatives flous pour surmonter certaines barrières du Lean Six Sigma liées aux phases d’analyse et d’amélioration de l’état actuel des processus de fabrication. Les approches proposées sont appliquées dans deux entreprise algériennes pour améliorer et contrôler la durabilité de leurs processus de fabrication.

With the evolution of the internet of things (IoT), and due to its significant need in the industry, Industrial IoT (IIoT) becomes the suitable naming for this accompaniment. IIoT changed the view of the industry intelligently and over the internet. This overlapping of IoT and industry requires special treatment when systems deal with heterogeneous devices in a distributed environment and complex tasks. In this paper, we propose a middleware solution based on multi-agents system (MAS) to handle the distributed control of complex systems autonomously in an industrial environment. The proposed middleware enables machine-to-machine (M2M) communications among the system’s components. In this work, we also addressed the distributed diagnostic for real industrial system using MAS with a new suitable communication strategy to support the heterogeneity and interoperability issued in IIoT and assure real-time monitoring and control. Finally, we present a qualitative evaluation of our solution on real case study (cement factory).

Cette thèse propose de concevoir et réaliser un système de reconnaissance automatique de la parole destiné à commander à distance un système audiovisuel à savoir : un Téléviseur. Le système global "bout en bout" se scinde en deux blocs : le premier cherche à extraire les meilleures caractéristiques à partir du signal vocal d’entrée. A cet effet, plusieurs techniques d’extraction de caractéristiques vont être examinées et testées. Concernant le deuxième bloc, nous mettons en évidence une multitude de techniques relevant du domaine de l’apprentissage profond, dont l’impact est d’adapter et de d’affirmer les caractéristiques extraites pour donner en final la classe de l’énoncé. La validation des différentes méthodologies présentées dans cette thèse a été effectuée sur la base de deux jeux de données réelles, le premier est tenu compte pour une évaluation initiale, tandis que le second est con\c cu exclusivement pour le système ASR proposé dans cette thèse. Les résultats obtenus ont certifié l’efficience des approches proposées. Le défi pour les travaux futurs est d’évaluer ce type de système dans des conditions plus réalistes avec des signaux vocaux issus des milieux bruités.

Modern wind turbines operate in continuously transient conditions, with varying speed, torque, and power based on the stochastic nature of the wind resource. This variability affects not only the operational performance of the wind power system, but can also affect its integrity under service conditions. Condition monitoring continues to play an important role in achieving reliable and economic operation of wind turbines. This paper reviews the current advances in wind turbine condition monitoring, ranging from conventional condition monitoring and signal processing tools to machine-learning-based condition monitoring and usage of big data mining for predictive maintenance. A systematic review is presented of signal-based and data-driven modeling methodologies using intelligent and machine learning approaches, with the view to providing a critical evaluation of the recent developments in this area, and their applications in diagnosis, prognosis, health assessment, and predictive maintenance of wind turbines and farms.

Since bearing deterioration patterns are difficult to collect from real, long lifetime scenarios, data-driven research has been directed towards recovering them by imposing accelerated life tests. Consequently, insufficiently recovered features due to rapid damage propagation seem more likely to lead to poorly generalized learning machines. Knowledge-driven learning comes as a solution by providing prior assumptions from transfer learning. Likewise, the absence of true labels was able to create inconsistency related problems between samples, and teacher-given label behaviors led to more ill-posed predictors. Therefore, in an attempt to overcome the incomplete, unlabeled data drawbacks, a new autoencoder has been designed as an additional source that could correlate inputs and labels by exploiting label information in a completely unsupervised learning scheme. Additionally, its stacked denoising version seems to more robustly be able to recover them for new unseen data. Due to the non-stationary and sequentially driven nature of samples, recovered representations have been fed into a transfer learning, convolutional, long–short-term memory neural network for further meaningful learning representations. The assessment procedures were benchmarked against recent methods under different training datasets. The obtained results led to more efficiency confirming the strength of the new learning path.

Condition Monitoring of photovoltaic systems plays an important role in maintenance interventions due to its ability to solve problems of loss of energy production revenue. Nowadays, machine learning-based failure diagnosis is becoming increasingly growing as an alternative to various difficult physical-based interpretations and the main pile foundation for condition monitoring. As a result, several methods with different learning paradigms (e.g. deep learning, transfer learning, reinforcement learning, ensemble learning, etc.) have been used to address different condition monitoring issues. Therefore, the aim of this paper is at least, to shed light on the most relevant work that has been done so far in the field of photovoltaic systems machine learning-based condition monitoring.

To ensure the continuity of electric power generation for photovoltaic systems, condition monitoring frameworks are subject to major enhancements. The continuous uniform delivery of electric power depends entirely on a well-designed condition maintenance program. A just-in-time task to deal with several naturally occurring faults can be correctly undertaken via the cooperation of effective detection, diagnosis, and prognostic analyses. Therefore, the present review first outlines different failure modes to which all photovoltaic systems are subjected, in addition to the essential integrated detection methods and technologies. Then, data-driven paradigms, and their contribution to solving this prediction problem, are also explored. Accordingly, this review primarily investigates the different learning architectures used (i.e., ordinary, hybrid, and ensemble) in relation to their learning frameworks (i.e., traditional and deep learning). It also discusses the extension of machine learning to knowledge-driven approaches, including generative models such as adversarial networks and transfer learning. Finally, this review provides insights into different works to highlight various operating conditions and different numbers and types of failures, and provides links to some publicly available datasets in the field. The clear organization of the abundant information on this subject may result in rigorous guidelines for the trends adopted in the future.