Scheduling with neural networks for flexible manufacturing systems

An application of neural networks in multiple-task scheduling problems is presented. The optimum task scheduling for manufacturing processes is, in general, an NP-complete problem for single server or manufacturing cell. The problem is more severe for scheduling many tasks with precedence constraints among them, timing requirements, set-up costs and completion deadlines to several manufacturing cells. The crossbar Hopfield network which is used to solve the classical traveling salesman problem is extended to a three-dimensional neuro-box network to solve multiple task scheduling on multiple servers. The complete formulation of the problem is presented. This includes the definition of the energy function for the neural network and the differential equations for the neurons. Several simulations are carried out and presented