Job-Shop Scheduling Using Genetic Algorithm and Reinforcement Learning

Which method schedules a job shop best - a classical dispatching rule, a reinforcement-learning dispatcher or a genetic algorithm? This project compares 7 Giffler-Thompson dispatching rules (SPT, MWKR, FIFO ...), a PPO dispatching policy trained on random 10x10 instances, three genetic algorithms (random keys as in MATLAB ga, POX with an active decoder, and a memetic GA with N5 tabu search) and Google OR-Tools CP-SAT as the exact reference on 138 standard benchmark instances (Fisher-Thompson, Lawrence, ABZ, ORB, Taillard; 36 to 2,000 operations). A web application schedules new instances and shows them as Gantt charts. GNU Octave runs the toolbox-free MATLAB code, and a complete MATLAB version (Global Optimization and Reinforcement Learning Toolboxes) is included.

Category: Industrial Engineering

Industrial Engineering Project IE01 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.