Heuristic Dynamic Scheduling Of One-of-a-kind Production Systems

Multiple product-types and small volume customized will be the main production mode in the twenty-first century. At the same time, manufacturing industry is in the process of becoming agile manufacturing and lean production. In this simulation, production planning and scheduling of production will be the key issue to make the manufacturing process efficiently and smoothly. It is very important to develop effective scheduling optimization algorithm to reduce cost and improve productivity. Thus, more and more researchers start pay their attention to this research.In this paper, we develop a new heuristic dynamic scheduling algorithm for one-of-a-kind production(OKP) system in which each product has its own tree-like process route and arrives one by one with exponentially distributed interval times. The aim of dynamic scheduling is to minimize the average sojourn time (i.e., the length of the time interval between arrival and departure) of all products in the system. To solve the problem, we develop the difference equations to capture its dynamics mathematically, in which the remained workload of all the products are state variables and the production rates of all the products are control variables. Based on this model, for each product at the OKP system, we calculate the fastest decreasing trajectory and slowest decreasing trajectory of its remained workload over time based on the technique of discrete event simulation on every scheduling point. Then we can construct the desired trajectories of the decreasing of remained workload according to the fastest and slowest decreasing trajectories for every product at the OKP system, and design a genetic algorithm-based method to trace the desired trajectories. To examine the efficiency of the heuristic dynamic scheduling algorithm, we construct some simulation experiments and comparing the proposed algorithm with some existing scheduling rules. The results of the simulation show that the performance of the proposed dynamic scheduling algorithm is much better than some existing scheduling rules. Finally, some managerial insights are summarized based on the results of simulation experiments.