Dynamic MPC-based scheduling in a smart manufacturing system problem

This paper introduces a dynamic scheduling algorithm designed to minimize makespan within a smart manufacturing system, accommodating delays in the production process. The proposed approach relies on Model Predictive Control (MPC) principles and adapts flow-shop scheduling theory to solve an open-shop scheduling problem. It aims to strike a balance between the ideal, delay-free solution and robustness in the case of processing time delays. By combining MPC theory with flow-shop scheduling, the algorithm offers a robust approach to open-shop scheduling problems, even with uncertain processing times. Iterated upon the arrival of each new job on the shop floor, the algorithm incorporates a control horizon to predict impending job arrivals and seamlessly integrates them into the scheduling process. Efficiency is examined through a comprehensive case study, where it is compared against a similar, offline scheduling algorithm. This novel method not only optimizes scheduling but also adapts to dynamic scenarios, reducing the computational demand and the information needed to optimize the production process, thus making it suitable for agile manufacturing environments. The results demonstrate the algorithm’s efficacy in achieving competitive scheduling performance with nearly the same makespan as the offline algorithm, while accounting for uncertainties in processing times. A robustness analysis confirms the reliability of the proposed approach, showing an average improvement of 5% in makespan across different delay magnitudes.