A fuzzy rule-based approach for dynamic control of kanbans in a generic kanban system

This paper studies the dynamic control of kanbans in a generic kanban system. A generic kanban system originating from the Toyota kanban system is a production control scheme in dynamic environments with variable demands and processing times. A dynamic kanban control approach is proposed to determine the number of kanbans used in the system. This approach includes two major functions: example extraction and fuzzy system generation. Examples are generated from simulation and desired examples are collected. Fuzzy rule-based systems are generated from desired examples to dynamically adjust the number of kanbans. Experiments are conducted to evaluate the proposed approach. The generated fuzzy systems are examined on the training process and its feasibility in general situations. This approach is also compared with the simulated annealing approach under the same system status. Both approaches yield similar system performance. The fuzzy system approach, however, is preferred because it does not require the prior demand information and it is a real-time control that can respond to new system status immediately.