Digital Twin of a Production Line Using Discrete-Event Simulation and Bayesian Optimisation

Where should a limited number of buffer places go in a production line whose machines break down at random? This project builds a digital twin of a five-machine line with finite buffers - a fast discrete-event simulator validated against an exact Markov chain, SimPy and the free simulator JaamSim - and compares equal allocation, a bottleneck rule and greedy search with Gaussian-process Bayesian optimisation on the true throughput of all 1,771 allocations. It adds preventive-maintenance planning, an event-log loop that re-estimates the machines' failure and repair times shift by shift and re-optimises, and a web application that answers what-if questions live. A complete MATLAB version is included.

Category: Industrial Engineering

Industrial Engineering Project IE03 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.