Papers by Philippe THOMAS
Just in Time (JIT) is a management philosophy that favorates improvement through inventory reduct... more Just in Time (JIT) is a management philosophy that favorates improvement through inventory reduction by producing only the right quantity at the right time. A kanban system is an important element of JIT philosophy. However, whereas kanban system is suitable when demand is stable, it should to be adjusted when demand is a significantly variable to be efficient. In addition, some works have proved that Petri nets are useful to model a kanban loop. Then, the objective of the present work is to exploit the analytical relation of Petri net models in order to improve the control of kanban system by using the real data provided by the system or by the product. This approach is based on the use of a continuous approximation of a discrete Petri net model in order to exploit a strategy of command based on gradient descent. In the present paper, two continuous approximations are proposed and compared.
Commande des feux de signalisation par réseaux de Petri hybrides
Journal Européen des Systèmes Automatisés, 2008
Des Carrefours «intelligents» Pour Une Meilleure Gestion Du Trafic Urbain
RÉSUMÉ. La gestion du trafic en milieu urbain nécessite le déploiement de solutions techniques in... more RÉSUMÉ. La gestion du trafic en milieu urbain nécessite le déploiement de solutions techniques innovantes pour améliorer la fluidité des déplacements et la qualité environnementale des transports de personnes et marchandises. Le laboratoire des Systèmes et Transports à Belfort en ...
en cause le concept de planification centralisée. Les nouvelles technologies de communication qui... more en cause le concept de planification centralisée. Les nouvelles technologies de communication qui commencent à prouver leurs capacités permettent d’envisager de nouvelles approches dans lesquelles le produit joue un rôle. Cependant diverses difficultés apparaissent alors pour le concepteur du nouveau système: définition des exigences, contraintes liées aux standards, coûts, etc. Ce papier se veut être un guide permettant d’aborder la mise en place de ce type de système de manière efficiente. Il présente dans un premier temps comment analyser le problème industriel, puis les différentes technologies envisageables actuellement ainsi que leurs avantages et inconvénients, et enfin comment prendre en compte le futur système en lui-même lors de sa conception. Nous finirons sur un aperçu des différents gains qui peuvent être tirés de la synchronisation des flux matières et informationnels en matière de décision réactive. Toutes les informations ont pu être illustrées dans le cadre d’une ap...
Proceedings of the 7th International Joint Conference on Computational Intelligence, 2015
In learning approaches for classification problem, the misclassification error types may have dif... more In learning approaches for classification problem, the misclassification error types may have different impacts. To take into account this notion of misclassification cost, cost sensitive learning algorithms have been proposed, in particular for the learning of multilayer perceptron. Moreover, data are often corrupted with outliers and in particular with label noise. To respond to this problem, robust criteria have been proposed to reduce the impact of these outliers on the accuracy of the classifier. This paper proposes to associate a cost sensitivity weight to a robust learning rule in order to take into account simultaneously these two problems. The proposed learning rule is tested and compared on a simulation example. The impact of the presence or absence of outliers is investigated. The influence of the costs is also studied. The results show that the using of conjoint cost sensitivity weight and robust criterion allows to improve the classifier accuracy.
Service Orientation in Holonic and Multi Agent Manufacturing and Robotics, 2013
The first objective of this chapter is to highlight some new Product Driven Systems (PDS) issues.... more The first objective of this chapter is to highlight some new Product Driven Systems (PDS) issues. Effectively, several possibilities have been proposed to give to products or objects capacities to react to environment modifications (especially in manufacturing and logistics context here). In particular, bioinspired approaches are now promising. All these new perspectives lead putting products in action according to collected information. That's why all technics leading to exploit and organize data are necessary. The main objective of the chapter is addressed in a second part, where we highlight why learning machines could be seen as a new way to transform data in useful knowledge.
Cet editorial introduit le numero special du Journal Europeen des Systemes Automatises, portant s... more Cet editorial introduit le numero special du Journal Europeen des Systemes Automatises, portant sur la gestion et l'impact de la qualite produit sur la production. Ce numero special regroupe des versions etendues d'articles selectionnes issus conjointement de la 10 th International Conference on Modeling, Optimization and SIMulation MOSIM'14 qui s'est deroule a Nancy (France) du 5 au 7 novembre 2014 et du 11 eme Congres International Pluridisciplinaire en Qualite, Surete de Fonctionnement et Developpement Durable QUALITA'15 qui s'est deroule a Nancy (France) du 17 au 19 mars 2015. Les papiers presentes dans ce numero special ont ete selectionnes sur la base de la qualite et de la diversite des propositions. La presentation de ces contributions sera precedee d'une courte introduction et d'etat de l'art.
Proceedings of the 13th International Joint Conference on Computational Intelligence, 2021
Predicting the set of lumbers which would be obtained from sawing a log at a specific sawmill is ... more Predicting the set of lumbers which would be obtained from sawing a log at a specific sawmill is a difficult problem, which complicates short and mid term decision making in this industry. While sawmill simulators able to simulate the sawing of a log from a 3D scan of its outer shape exist, they can be extremely computationally intensive. Several alternative approaches based on machine learning algorithms and different set of features were explored in previous works. This paper proposes the use of one hidden layer perceptrons, and a vector of features build from dissimilarities from the scans to a set of selected wood logs, chosen as the class medoids. Several architectures are tested and compared to validate the pertinence of the proposed set of medoid-based features. The lowest mean squared error was obtained for MISO neural networks with a sigmoid output activation function, to constrain the output value ranges.
2019 International Joint Conference on Neural Networks (IJCNN), 2019
Predict the indoor air quality becomes a global public health issue. That's why Airbox lab® compa... more Predict the indoor air quality becomes a global public health issue. That's why Airbox lab® company develops a smart connected object able to measure different physical parameters including concentration of pollutants (volatile organic compounds, carbon dioxide and fine particles). This smart object must embed prediction capacities in order to avoid the exceedance of an air quality threshold. This task is performed by neural network models. However, when some events occur (change of people's behaviors, change of place of the smart connected object as example), the embedded neural models become less accurate. So a relearning step is needed in order to refit the models. This relearning must be performed by the smart connected object, and therefore, it must use the less computing time as possible. To do that, this paper propose to combine a control chart in order to limit the frequency of relearning, and to compare three learning algorithms (backpropagation, Levenberg-Marquardt, neural network with random weights) in order to choose the more adapted to this situation.
Computers in Industry, 2018
In recent times, the manufacturing processes are faced with many external or internal (the increa... more In recent times, the manufacturing processes are faced with many external or internal (the increase of customized product rescheduling , process reliability,..) changes. Therefore, monitoring and quality management activities for these manufacturing processes are difficult. Thus, the managers need more proactive approaches to deal with this variability. In this study, a proactive quality monitoring and control approach based on classifiers to predict defect occurrences and provide optimal values for factors critical to the quality processes is proposed. In a previous work (Noyel et al. 2013), the classification approach had been used in order to improve the quality of a lacquering process at a company plant; the results obtained are promising, but the accuracy of the classification model used needs to be improved. One way to achieve this is to construct a committee of classifiers (referred to as an ensemble) to obtain a better predictive model than its constituent models. However, the selection of the best classification methods and the construction of the final ensemble still poses a challenging issue. In this study, we focus and analyze the impact of the choice of classifier types on the accuracy of the classifier ensemble; in addition, we explore the effects of the selection criterion and fusion process on the ensemble accuracy as well. Several fusion scenarios were tested and compared based on a real-world case. Our results show that using an ensemble classification leads to an increase in the accuracy of the classifier models. Consequently, the monitoring and control of the considered real-world case can be improved.
Computers in Industry, 2016
HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Journal Européen des Systèmes Automatisés, 2016
HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Limiter l'impact de la non-qualité sur la gestion des flux de production dans les ateliers à forts taux de reprises. Application au cas d'un fabricant de mobiliers laqués de très haute qualité
Neural Processing Letters, 2014
Optimizing the structure of neural networks remains a hard task. If too small, the architecture d... more Optimizing the structure of neural networks remains a hard task. If too small, the architecture does not allow for proper learning from the data, whereas if the structure is too large, learning leads to the well-known overfitting problem. This paper considers this issue, and proposes a new pruning approach to determine the optimal structure. Our algorithm is based on variance sensitivity analysis, and prunes the different types of unit (hidden neurons, inputs, and weights) sequentially. The stop criterion is based on a performance evaluation of the network results from both the learning and validation datasets. Four variants of this algorithm are proposed. These variants use two different estimators of the variance. They are tested and compared with four classical algorithms on three classification and three regression problems. The results show that the proposed algorithms outperform the classical approaches in terms of both computational time and accuracy.
Retour D'Expérience Industrielle Sur Le Choix D'Une Technologie D'Information Portées Par Les Produits
RÉSUMÉ : Les problèmes actuels rencontrés dans les entreprises amènent la communauté scientifique... more RÉSUMÉ : Les problèmes actuels rencontrés dans les entreprises amènent la communauté scientifique à remettre en cause le concept de planification centralisée. Les nouvelles technologies de communication qui commencent à prouver leurs capacités permettent d'envisager de nouvelles approches dans lesquelles le produit joue un rôle. Cependant diverses difficultés apparaissent alors pour le concepteur du nouveau système : définition des exigences, contraintes liées aux standards, coûts, etc. Ce papier se veut être un guide permettant d'aborder la mise en place de ce type de système de manière efficiente. Il présente dans un premier temps comment analyser le problème industriel, puis les différentes technologies envisageables actuellement ainsi que leurs avantages et inconvénients, et enfin comment prendre en compte le futur système en lui-même lors de sa conception. Nous finirons sur un aperçu des différents gains qui peuvent être tirés de la synchronisation des flux matières et ...
Implantation of an on-line quality process monitoring
Because product quality level is today a key factor for companies' competitiveness and becaus... more Because product quality level is today a key factor for companies' competitiveness and because it is really hard to control the manufacturing systems, there is a lot of scientific methods implemented on the shop-floor to improve quality at workstations. In our previous works, we have shown that the on-line quality control assisted by Neural Network seems to be a good alternative to the well-known industrial methods such as Taguchi Method. Nevertheless we have highlighted that a drift appears between the model and the process reality. In this paper, we propose a way to resynchronize the on-line control system behavior with it. This approach allows us to assure that the model stay robust and adaptable for the quality prediction. We will illustrate this with the Acta-Mobilier case, which is a high quality lacquerer company in the furniture industry.
CART for Supply Chain Simulation Models Reduction
IFIP Advances in Information and Communication Technology, 2014
IFAC Proceedings Volumes, 2005
This paper is about the flow control design of hybrid dynamical systems modelled with continuous ... more This paper is about the flow control design of hybrid dynamical systems modelled with continuous Petri nets and described as a set of bilinear state space representations. The marking vector stands for the state space vector, where some places are observable and other are not. The transitions are divided into controllable ones and uncontrollable ones. Gradient-based controllers are investigated in order to adapt the firing speeds of the controllable transitions according to a desired trajectory of the marking. The equilibriums and stability of the controlled system are discussed according to the controller time parameter.
Neurocomputing, 1997
The problem of non-linear Single Input Single Output system identification in the presence of lar... more The problem of non-linear Single Input Single Output system identification in the presence of large errors in data is considered. Combining the capabilities of neural networks to solve non-linear problems by learning and a robust recursive prediction error learning rule based on the modeling of the errors, a new algorithm is drawn up. Its potential is illustrated through simulation studies.
IFAC Proceedings Volumes, 2012
The first objective of this paper is to highlight some new Product Driven Systems (PDS) issues. E... more The first objective of this paper is to highlight some new Product Driven Systems (PDS) issues. Effectively, several possibilities have been proposed to give to products or objects capacities to react to environment modifications (especially in manufacturing and logistics context here). In particular, bio-inspired approaches are now promising. All these new perspectives lead putting products in action according to collected information. That's why all technics leading to exploit and organize data are necessary. The main objective of the paper is addressed in a second part, where we highlight why learning machines could be seen as a new way to do that.
Uploads
Papers by Philippe THOMAS