Study On Make-to-Order Dynamic Production Scheduling Problem With Material Resources Constraint

With global manufacturing developing into intelligent manufacturing,it is more obvious to arrange production according to customers’ needs which become more and more diverse and characteristic.In order to satisfy customers’ needs,companies gradually change traditional make-to-stock production into make-to-order production.Under the background,material resources constraint is considered in the paper.And scheduling is made with considering resources supply and production as a whole.Because on-time delivery is an important service standard,the problem is make-toorder dynamic production scheduling problem under material resources constraint with the function objective of minimizing total tardiness.Core manufacturing company accepts all stochastic orders and sends information of orders to the warehouse.Then,it orders material resources based on inventory from upstream suppliers.Lastly,a schedule is made based on material resources supply.Due to uncertainty of orders,the problem is dynamic.When dynamic event(acceptance of a new order)happens,unfinished products and ones from the new order are rescheduled based on current production status.First of all,the problem is analyzed and reasonable assumptions are proposed.The function objective is to minimize total tardiness.Mathematical constraints are proposed and explained.A mixed integer programming model is presented.Furthermore,two improved metaheuristics,Modified Artificial Immune System algorithm(MAIS)and Multiple Variable Neighborhood Search(MVNS),are proposed.Based on the understanding of design idea and basic theory,the artificial immune system algorithm is modified by learning advantages of natural immune remembering and processing information.MAIS is composed of V(D)J Genetic Recombination,Somatic Hypermatic,Isotype Switching and Second Immune Response.Multiple Variable Neighborhood Search algorithm is composed of Initialization,Four Neighborhood Structures,Disturbing,Neighborhood Switching and Strengthening.Neighborhood Switching makes it self-revise to jump out of the local optimal solution.Strengthening accelerates its convergence.Disturbing makes local solution becomes better when local optimal solution has not been improved continuously.Finally,computational experiments are conducted to analyze and evaluate performance of the algorithems.240 problems are generated by combining different problem’s parameters.All algorithms are coded in C++ language.Comparisons are made by relative percentage deviation,computational times,standard deviation and statistical analysis.Computational experiments prove that Modified Artificial Immune System algorithm and Multiple Variable Neighborhood Search can solve the problem and obtain nearly optimal solution in a short time.