CLOUD-BASED ADAPTIVE SHOP-FLOOR SCHEDULING CONSIDERING MACHINE TOOL AVAILABILITY

Short-term scheduling belongs to the typical decision making problems in manufacturing that continue to draw attention from industry and academia due to its inherent difficulties. The trend of mass customization and the increasing product variety generate further uncertainties and turbulences on modern shop-floors, thus, making scheduling a challenging daily problem. These challenges dictate the need for replacing rigid centralized scheduling tools with adaptive and robust scheduling solutions. The integration between ICT based decision support tools in manufacturing can be further enhanced to achieve shop-floor awareness and a common information flow, which is necessary to improve decision-making. Towards this objective, advanced monitoring techniques consisting of smart sensor networks and seamless communication procedures can provide the required awareness to decision making process. Further to that, Cloud, as an emerging enabling technology, can support the integration among multiple IT tools and provide ubiquitous access to data. Cloud-based and resource-aware scheduling tools are therefore considered as enablers for increasing the adaptability and agility of a manufacturing system. The proposed research work, presents a cloud-based framework consisting of a monitoring service and a short-term scheduling application that aims to generate and dispatch feasible and highly-productive schedules in a timely manner. The short-term scheduling application is enriched with data obtained by the monitoring service and generates resource-aware schedules by considering not only machine tools suitability but also their imminent status and availability. The scheduling application utilizes an intelligent search algorithm, which allows the generation of alternative schedules and their evaluation though a set of multiple conflicting criteria including among others cost, time and quality. The produced schedules are assessed using a set of performance indicators of makespan and resource utilization. The monitoring service gathers data from two data sources, namely a multi-sensory system and the machine tool operator. Through an information fusion procedure, the monitoring service provides to the scheduling application the machine tools status as well as the machine tools available time windows. The sensory system is deployed on five axes work-centers to monitor the axis and spindle drives in near real-time. The human operator reports to the monitoring system the status of the machine tool, the currently running task, and the cutting-tool availability through mobile devices on the shop-floor. The information fusion technique, consisting of the Analytic Hierarchy Process and the Dempster's Shafer theory of evidence, processes these heterogeneous information sources and derives the status of the machine tool and future availability windows. The proposed framework is applied and validated in a real-life case study obtained from a high precision mold-making industry.