Scheduler With Buffer and Transportation Constraints for Inserting Rush Orders Without Deadlocks During Agent Convergence

Abstract

Predictability is essential in modern industry for maintaining effective management and scaling production more precisely. Scheduling algorithms capable of obtaining the best solutions to problems with buffer and transportation constraints are crucial for predictability. This paper presents a scheduling algorithm for predictive scheduling problems with buffer and transportation constraints that can use up to five methods to obtain event matrices from a production model. The matrices were resized to avoid repeating resources in parallel events during the synchronization periods of the system. The algorithm selects the matrix with the lowest scaling cost to generate the production schedule. The schedule is deadlock-free owing to buffer insufficiency and overlay, thus keeping the resource allocation and task sequencing consistent with the initial model. The scheduling algorithm is suitable for supporting two types of multiagent consensus to guarantee the execution of regular and urgent schedules in a flexible manufacturing system. Predefined finite-time consensus handles regular-order production schedules, while a group or cluster consensus guarantees the execution of rush-order schedules. A method based on time windows is used to validate when rush orders are inserted into the system such that no deadlocks occur owing to buffer limitations and the overlapping of tasks on resources occupied by regular order schedules. Experiments were conducted to evaluate the proposed algorithms, confirming that the approach increases the flexible manufacturing system’s ability to enter rush orders without rescheduling by 42.86%.

Keywords

• Multiagents
• flexible manufacturing systems
• predictive scheduling
• rush orders
• deadlocks