۱- Introduction

۲- Related works

۳- Ramp-up Decision Support System

۴- Experimental procedure and results

۵- Discussion

۶- Conclusions

References

Abstract

Production ramp-up is a key phase during the introduction or changeover of a production system. Process calibration and tuning are inevitably required to make such a system fully operational and let it reach its maximum production yield. A complex decision-making process takes place in order to optimally tune the system and requires a long time for testing and experimenting that will determine the system behaviour. This work considers the sequential nature of ramp-up and proposes a Cyber-Physical Systems approach based on data capturing, learning mechanisms and knowledge extraction, leading to an Industry 4.0 compliant Decision Support System (DSS) for human operators. The proposed system is implemented as an online DSS and also supports offline learning using previously gathered knowledge. A number of experiments have been carried out on a micro scale assembly station, validating the expected benefits of the proposed DSS. Results show a reduction of over 40% in the number of ramp-up steps required when using the DSS.

Introduction

Nowadays, production systems (PSs) are subject to rapid technological changes, which led to very short product lifecycles and high variety of products, to meet the growing demand of product customization (Kalir and Rozen, 2018). In addition to that, PSs becomes more modular and able to change from one configuration to another (Colledani et al., 2018). The rapid introduction of new products, mostly complex, and changeover cause a prolonged and more frequent production ramp-up (Letmathe and Rößler, 2019). Ramp-up phase is very important for PSs because having high outputs with good quality and fast cycle time are the key factors for increasing profits, where at the beginning of product life-cycle product prices are the heights and competition is scarce (Kalir and Rozen, 2018). Production ramp-up is a decision-making process where human experts decide on the best actions to fine-tune the process. It is a highly complex parameter tuning process with many interrelated factors leading to a well-defined goal. Studies on ramp-up report long lead times of up to a few months from the initial sys-tem built to full production (Terwiesch and Bohn, 2001). Since the product, PS, and the supply chain are new, uncertainty is very high, making the ramp-up process very difficult to manage (Hansen and Grunow, 2015). Also, ramp-up process is unstable, expressed in the unpredictability of the system at this stage, which makes planning proactive steps to avoid some problems ineffective for many ramp-up problems (Schuh et al., 2015).