The use of threshold logic to improve performance in scheduling a manufacturing process

Abstract

Simulation and threshold logic are used to demonstrate the feasibility of a learning system that increases production output. Production is simulated with a discrete, next-event model having multiple paths and multiple processes on each path. Switches located throughout the model can direct production units to avoid bottlenecks and improve output. Each switch in the model is controlled by the action of threshold gates. For the process time configurations tested, minima were found to be associated with one or more of three threshold values set in 92% of the configurations. Using the three threshold value sets and a rule base for process time configurations that are not optimized by one of the three sets, a system may be created that avoids the computational costs and time delays of a multiple-trial optimizing system.<>