Entity

Time filter

Source Type

Mount Lebanon, Lebanon

Neamatian Monemi R.,University of Lille Nord de France | Danach K.,University of Lille Nord de France | Danach K.,University of Artois | Danach K.,Islamic University of Lebanon | And 5 more authors.
Expert Systems with Applications | Year: 2015

We take into account a parallel heterogenous machine scheduling problem arising in maintenance planning of heterogeneous wells. This problem particularly arises in the context of workover rig scheduling. The oil wells need regular maintenance to ensure an optimal level of production. After oil production being decreased at some wells, appropriate workover rigs with compatible service capacity, are deployed to serve the wells at discrete locations. Every well needs a certain level of maintenance and rehabilitation services that can only be offered by compatible workover rigs. A new mixed integer linear programming model is propose for this problem that is an arc-time-indexed formulation. We propose a heuristic selection type hyper-heuristic algorithm, which is guided by a learning mechanism resulting in a clever choice of moves in the space of heuristics that are applied to solve the problem. The output is then used to warm start a branch, price and cut algorithm. Our numerical experiments are conducted on instances of a case study of Petrobras, the Brazilian National Petroleum Corporation. The computational experiments prove the efficiency of our hyper-heuristic in searching the right part of the search space using the right alternation among different heuristics and confirms the high quality of solutions obtained by our hyper-heuristic. © 2015 Elsevier Ltd. All rights reserved. Source


Mcheick H.,University of Quebec at Chicoutimi | Mohammed Z.R.,Islamic University of Lebanon | Lakiss A.,Lebanese University
Proceedings - 2011 9th International Conference on Software Engineering Research, Management and Applications, SERA 2011 | Year: 2011

The Information technology Infrastructure plays an important role in the success of business applications. However, these applications suffer from performance and availability. In this vain, resource utilization is out of balance. Load balancing is very important approach to minimize the execution time because it has many processes units that are running in the same time. It is important to decompose the tasks among processors to achieve load balance. We distinguish two approaches to solve load balance: static and dynamic. Each approach has many algorithms which are not yet evaluated to understand the advancements as well as weaknesses over each other. The main purpose of this paper is to help in design of new algorithms in future by studying the behavior of various existing algorithms. We are going to evaluate these algorithms based on new parameters such as process migration, overhead, scalability and availability. © 2011 IEEE. Source


Danach K.,Islamic University of Lebanon | Khalil W.,Lebanese University | Gelareh S.,University of Artois
2015 3rd International Conference on Technological Advances in Electrical, Electronics and Computer Engineering, TAEECE 2015 | Year: 2015

The service network design problems arising in liner shipping industry are very intractable problems. Several exact method are proposed for such problems where almost all of them are limited by the instance size that can be resolved. In this article, we consider the problem of designing multiple strings among a set of ports, in order to maximize the industry profit. In this work, we develop hyper-heuristics by proposing different low level heuristics categorized as constructive, improvement, perturbation etc. The low level heuristics are guided by a meta-heuristic algorithm that is supported by data mining techniques to attain balancing between intensification and diversification strategies in choosing the best heuristics series to be applied. © 2015 IEEE. Source


Kader I.A.,Islamic University of Lebanon
AIP Conference Proceedings | Year: 2013

In this paper, we analyze the bounds of path number in a directed graph, especially in the tournament Tn=(X,U), where we prove that: P(T n)≤n24-2 and from this remarkable result, we give some subclasses of tournaments for which we have P(Tn)=e(T n)=n24-2 (i.e. lower bound of P(G)=upper bound of P(G)) especially if An=(X,U) is the tournament having exactly m-n+1 elementary circuits we have: P(An)=e(An)=n24-2. Also we give a general theorem concerning the classes of directed graphs satisfying P(G)=e(G). The result obtained: 1. A procedure that allows a directed graph of order n satisfying P(G)=e(G), another class of directed graph G1 of order n+1 such that P(G1)=e(G1). 2. Generalized certain results proved by Alspach and Ore concerning the path partition number in directed graphs. © 2013 AIP Publishing LLC. Source


Barakat M.,University of Le Havre | Barakat M.,Lebanese University | Lefebvre D.,University of Le Havre | Khalil M.,Lebanese University | And 2 more authors.
International Journal of Machine Learning and Cybernetics | Year: 2013

Neural networks have been widely used in the field of intelligent information processing such as classification, clustering, prediction, and recognition. In this paper, a non-parametric supervised classifier based on neural networks is proposed for diagnosis issues. A parameter selection with self adaptive growing neural network (SAGNN) is developed for automatic fault detection and diagnosis in industrial environments. The growing and adaptive skill of SAGNN allows it to change its size and structure according to the training data. An advanced parameter selection criterion is embedded in SAGNN algorithm based on the computed performance rate of training samples. This approach (1) improves classification results in comparison to recent works, (2) achieves more optimization at both stages preprocessing and classification stage, (3) facilitates data visualization and data understanding, (4) reduces the measurement and storage requirements and (5) reduces training and time consumption. In growing stage, neurons are added to hidden subspaces of SAGNN while its competitive learning is an adaptive process in which neurons become more sensitive to different input patterns. The proposed classifier is applied to classify experimental machinery faults of rotary elements and to detect and diagnose disturbances in chemical plant. Classification results are analyzed, explained and compared with various non-parametric supervised neural networks that have been widely investigated for fault diagnosis. © 2012 Springer-Verlag. Source

Discover hidden collaborations