Optimizing operations of flexible assembly systems: demonstration of a digital twin concept with optimized planning and control, sensors and visualization

Abstract

This paper presents the development of an optimized planning and control method for flexible manufacturing and assembly systems. While the significant potential of flexible manufacturing concepts to help producers adapt to market developments is recognized, the complexity of the flexible systems and the need to optimally plan and control them is a major obstacle in their practical implementation. Thus, this paper aims to develop a comprehensive digital planning method, based on a digital twin and to demonstrate the feasibility of the approach for practical application scenarios. The approach consists of four modules: (1) a simulation-based optimization module that applies reinforcement learning and genetic algorithms to optimize the module configuration and job routing in cellular reconfigurable manufacturing systems; (2) a synchronization module that links the physical and virtual systems via sensors and event handling; (3) a sensor module that enables a continuous status update for the digital twin; and (4) a visualization module that communicates the optimized plans and control measures to the shop floor staff. The demonstrator implementation and evaluation are implemented in a learning factory. The results include solutions for the method components and demonstrate their successful interaction in a digital twin, while also pointing towards the current technology readiness and future work required to transfer this demonstrator implementation to a full-scale industrial implementation.