Estimating the Throughput of a Closed-Queuing Network Production System Using Machine Learning

Any production system comprising machines and material handling devices is prone to queuing effects and other uncertainties due to reliability issues. An estimate of the long-run throughput rate of the system is therefore important for release decisions in production planning. Within the Industry 5.0 context, there is frequent and real-time data capture through sensor technologies enabling the monitoring and control of such systems. In this paper, a Machine Learning approach is proposed to estimate the throughput of a system comprising unreliable machines with non-stationary processing times and an automated material handling device. We validate our approach through a context-based simulator that uses the Mean-Value Approximation for closed-network queuing systems.