Job Flow Control for Buffers in a Job-Shop Manufacturing System

This paper investigates the job flow control problem for a network of buffers in a job-shop manufacturing system, where jobs are stored in buffers in order to supply a job demand to the job shop. For each buffer, a job flow control law is proposed, which requires that each buffer has access to the average total number of jobs stored in all buffers. It is shown that the control objectives of balancing the stock levels of the buffers and delivering the desired total job demand are both achieved if the proposed job flow control law is implemented for each buffer. Moreover, to relax the requirement of each buffer having access to the average total number of jobs among all buffers, two types of dynamic average consensus algorithms acting as distributed estimators are constructed for each buffer, each of which can provide an estimate of the average total number of stored jobs. By leveraging either type of distributed estimators, the proposed job flow control law is implemented in a distributed fashion. Simulation studies are provided to verify the effectiveness of the proposed control law.