Application Of Artificial Neural Network To Job-Shop Schedule

Manufacturing Scheduling is an important but difficult task, and is often formulated as a combinationial optimization problem. To effectively solve this class of problems, a novel optimization method has recently been developed by combining neural network optimization. In general, Job-Shop schedule is that available manchine aggregate proceed distributation with task.to satisfy performance requirement.This paper is to explore architectural design issues for the hardware implementation of such neural networks for Job-Shop scheduling. Hopfield-type neural networks were developed for unconstrained optimization based on the "Lyapunov stability theory" of dynamic systems: if a network is "stable", its "energy"will decrease to a minimum as the system approaches its "equilibrium state". If one can properly set up a network that maps the objective function of an optimization problem onto an "energy function", then the solution is a natural result of network convergence, and can be obtained at a very fast speed.For constrained optimization, the Hopfield-type networks convert a constrained problem to an unconstrained one by having penalty terms on constraint violations. A tradeoff between solution optimality and constraint satisfaction has to be made through the fine turning of penalty coefficients. The tradeoff, however, is generally difficult to make. In addition, Hopfield-type networks may possess many local minima. Since escaping form local minima is not an easy task, the solution quality depends highy on initial conditions.A general introduction to the thesis:1. An introducton of the theories of neural network applied to the JSP2. An applicative analysis of HNN in solving the JSP3. The improvement of HNN structure and present a three-layer HNN struture.4. To realize intelligent algorithms library management system by using BP neural network...