An artificial neural network optimized by a genetic algorithm for real-time flow-shop scheduling

A job-shop scheduling method using a three-layered neural network optimized by a generic algorithm, which was called a GANN scheduling method was a flexible and practical quasi-optimal scheduling method. However, further improvements of the present GANN scheduling system are required for rapid flow-shop rescheduling in chemical processes for multipurpose production. In this study we investigated the effect of improvements of the GANN scheduling system on the efficiency of rescheduling when new jobs were appended in a chemical process with some buffer tanks. In results, the former GANN scheduling method could be developed to a practical real-time scheduling system for process problems.