Manufacturing Multiagent System for Scheduling Optimization of Production Tasks Using Dynamic Genetic Algorithms

Abstract

This work faces a common yet difficult area of scheduling: that of distributing jobs to human operators that expose different production behaviors within a production line. The use of multiagent systems and genetic algorithms (GAs) is proposed to solve this type of problem. We suggest a system whose main components are avatar agents in charge of representing each human operator and a scheduler agent in charge of scheduling by the use of GAs. The system developed proved advantageous in the simulations experiments, reaching an average increase of 8.25% in the production rate, 59% decrease in the average of the operators' idleness, and 83% decrease in the standard deviation of the operators' idleness. Though quality improved only an average of 0.44%, the result may be deemed important in view of the high quality of the production line used as benchmark.