Description

Abstract

    The proposed modification consists in a suitable choice of the NN (Neural Network)

    initialization starting time of the JSSP (Job Shop Scheduling Problem). This optimal choice draws a fast training of the neural network and near-optimal solution.

    Hence, an iterative neural network is used in this paper describes some basic concepts and

    notations of JSSP (Job Shop Scheduling Problem). The same section reviews the integer linear

    programming representation with mathematical formulation.

     It gives the resolution of the integer linear programming representation for a job-shop scheduling

p  problem using NN (Neural Network).

      The neural unit model, the connection between the neural constraints units, and their architecture

(   sequence  and resource constraints) are then defined.

   We propose a new initialization method that is other than a Heuristic, based directly on a

database of a problem of a classic Job-shop scheduling problem.