Souhail Dhouib

Adaptive record-to-record travel method to solve lexicographic goal programming models

In this paper, a record-to-record travel (RRT) algorithm with an adaptive memory named taboo central memory (TCM) is adapted to solve the lexicographic goal programming problem. The proposed method can be applied to non-linear, linear, integer and combinatorial goal programmes. Because that the RRT has no memory, the adaptive memory TCM is inserted to diversify research. Computational experiments in several types of problems with different variable types (integer, continuous, zero-one and discrete) collected from the literature demonstrate that the proposed metaheuristic reaches high-quality solutions in short computational times. Furthermore, it requires very few user-defined parameters.

A Record to Record Travel metaheuristic to minimize total dispersion in food industry

In this paper a Record to Record Travel (RRT) metaheuristic is proposed to minimize the manufacturing batch dispersion in order to optimise traceability in food industry. Computational results on sausage manufacturing in a French food company sets show the effectiveness and the high performance of the proposed RRT approach. In comparison to the performance of previous genetic algorithm, the proposed one was found superior. Furthermore, it requires very few user-defined parameters.

Artificial bee colony metaheuristic to find pareto optimal solutions set for engineering design problems

In this paper, an Artificial Bee Colony (ABC) metaheuristic is adapted to find Pareto optimal solutions set for Goal Programming (GP) Problems. At first, the GP model is converted to a multi-objective optimization problem (MOO) of minimizing deviations from fixed goals. At second, the ABC is personalized to support the MOO by means of a weighted sum formulation for the objective function: solving several scalarization of the objective function according to a weight vector with non-negative components. The efficiency of the proposed approach is demonstrated by nonlinear engineering design problems. In all problems, multiple solutions to the goal programming problem are found in short computational time using very few user-defined parameters.

Goal programming using multiple objective hybrid metaheuristic algorithm

In this paper, a Goal Programming (GP) model is converted into a multi-objective optimization problem (MOO) of minimizing deviations from fixed goals. To solve the resulting MOO problem, a hybrid metaheuristic with two steps is proposed to find the Pareto sets solutions. First, a Record-to-Record Travel with an adaptive memory is used to find first non-dominated Pareto frontier solutions preemptively. Second, a Variable Neighbour Search technique with three transformation types is used to intensify every non dominated solution found in the first Pareto frontier to produce the final Pareto frontier solutions. The efficiency of the proposed approach is demonstrated by solving two nonlinear GP test problems and three engineering design problems. In all problems, multiple solutions to the GP problem are found in one single simulation run. The results prove that the proposed algorithm is robust, fast and simply structured, and manages to find high-quality solutions in short computational times by efficiently alternating search diversification and intensification using very few user-defined parameters.

A multi Start adaptive Variable Neighborhood Search metaheuristic for the aircraft landing problem

In this paper, a variable neighborhood search metaheuristic is enriched with a multi start technique and an adaptive taboo memory: the SVNS metaheuristic. In the proposed SVNS method, the diversification phase is ameliorated by launching the VNS metaheuristic in multi start technique and by archiving each blocked solution in taboo memory with a three different shames. This taboo memory governs the selection move for the neighborhood. The selection is performed in the neighborhood of each current solution. The SVNS metaheuristic is used to optimize the multiple runways aircraft landing problem. Computational experiments in several problems collected from the literature, instances from OR-Library, demonstrate that the proposed SVNS metaheuristic reaches high-quality solutions using very few user-defined parameters: the Kruskal-Wallis statistic test is used to prove that.

A variable neighbourhood search method to solve hierarchically engineering design problems

In this paper, a Variable Neighborhood Search (VNS) method is used to solve hierarchically engineering design problems. This method is adapted to solve problems with any type of variables, i.e. integer, discrete and continuous variables. In the proposed method, a decent algorithm with an adaptive memory is used to perform solutions by three systematic transformations techniques. This method allows finding high‐quality solutions in short computational time. Computational results prove that, and demonstrate that the proposed Hierarchical VNS (HR‐VNS) method performs all methods presented in the literature for engineering design problems.

Optimising Case Study Personnel Scheduling Problem Using an Artificial Bee Colony Algorithm

The population-based meta-heuristics are usually inspired from nature and unlike to the local search meta-heuristics which bring in a unique solution, these algorithms are able to manipulate a group of acceptable solutions at each stage of the research process. Today, the population-based meta-heuristics are widely used for the optimisation of NP-hard problems. Among these meta-heuristics, the algorithm of the artificial bee colony ABC, which is inspired by the forage behaviour of the honey bee in nature. In this paper, we suggest to solve an NP-hard problem, consisting in seafaring staff scheduling within a Tunisian company by means of the ABC algorithm. The objective of this work is to provide the company with schedules guaranteeing improved staff rest levelling compared with that traditionally applied. Besides, the obtained results show the performance of the ABC algorithm in the improvement of the cover rate for the scheduling with regard to the already proposed methods in literature.

Enriched artificial bee colony metaheuristic for hierarchical goal programming engineering design problems

In this paper, a standard artificial bee colony ABC metaheuristic was enriched to optimise hierarchical goal programming GP engineering design problems. The proposed method was first adapted to support the GP by means of minimising deviations from fixed goals, and then enriched by the great deluge metaheuristic, especially in the onlooker bee phase, in order to increase its local search ability. The proposed metaheuristic is called great deluge artificial bee colony GD-ABC. Experiments on four engineering design problems showed that the proposed GD-ABC metaheuristic converges rapidly and efficiently to solve those problems. The nonparametric Mann-Whitney U test was used to prove the comparison.

Upper bounds for time and accuracy improvement of dynamic time warping approximation

Dynamic time warping (DTW) consists at finding the best alignment between two time series. It was introduced into pattern recognition and data mining, including many tasks for time series such as clustering and classification. DTW has a quadratic time complexity. Several methods have been proposed to speed up its computation. In this paper, we propose a new variant of DTW called dynamic warping window (DWW). It gives a good approximation of DTW in a competitive CPU time. The accuracy of DWW was evaluated to prove its efficiency. Then the KNN classification was applied for several distance measures (dynamic time warping, derivative dynamic time warping, fast dynamic time warping and DWW). Results show that DWW gives a good compromise between computational speed and accuracy of KNN classification.

Artificial Bee Colony to solve seafaring staff scheduling problem: A real case

Personnel scheduling problem is a combinatorial optimization problem which belongs to NP-hard set of problems. In this paper, we propose to solve a seafaring staff scheduling problem through the use of the Artificial Bee Colony method. The central aim of this work is to offer a productive way in order to release a benchmarking between the results produced by the artificial Bee Colony method and those obtained by the traditionally applied method. The results show that this method outperforms the traditional one. Besides, this algorithm enables us to give the maximum of coverage with a less computational time.