Browsing by Author "Hemmak, Allaoua"

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemMetadata only
    Combination of Genetic Algorithm with Dynamic Programming for Solving TSP
    (Université de M'sila, 2017-07-02) Hemmak, Allaoua
    This paper presents a combination of Genetic Algorithm (GA) with Dynamic Programming (DP) to solve the well-known Travelling Salesman Problem (TSP). In this work, DP is integrated as a GA operator with a certain probability. In specific, at a given GA generation, the individuals are subdivided into a number of equal segments of genes, and the shortest path on each segment is obtained by applying a DP algorithm. Since the computational complexity of the DP is O (k22k), it becomes of O(1) when k is small. Experimental analyses are conducted to investigate the impact and trade-offs among DP probability, segment size and time processing on the solution quality and computational effort. In addition, we will implement a basic GA approach to compare results and show the contribution of combination of combination approach. Experimental results on benchmark instances showed that the combined GA-DP algorithm reduces significantly the computational effort, produces a clearly improved solution quality and avoids early premature convergence of GA.
  • Thumbnail Image
    ItemMetadata only
    New Properties for Solving the Single- Machine Scheduling Problem with Early/Tardy Jobs
    (Université de M'sila, 2016-07-12) Hemmak, Allaoua; Bouderah, Brahim
    This paper presents a mathematically enhanced genetic algorithm (MEGA) using the mathematical properties of the single-machine scheduling of multiple jobs with a common due date. The objective of the problem is to minimize the sum of earliness and tardiness penalty costs in order to encourage the completion time of each job as close as possible to the common due date. The importance of the problem is derived from its NP-hardness and its ideal modeling of just-in-time concept. This philosophy becomes very significant in modern manufacturing and service systems, where policy makers emphasize that a job should be completed as close as possible to its due date. That is to avoid inventory costs and loss of customer’s goodwill. Five mathematical properties are identified and integrated into a genetic algorithm search process to avoid premature convergence, reduce computational effort, and produce high-quality solutions. The computational results demonstrate the significant impact of the introduced properties on the efficiency and effectiveness of MEGA and its competitiveness to state-of-the-art approaches.