چکیده:
Different inventory control systems try to determine how much and when to order at the least relevant cost while maintaining a desirable service level for customers. In this article, a continuous review stochastic inventory system, with three objectives, is optimized. In this model, contrary to the traditional inventory models, customer service is not considered a shortage cost in the objective function. But the frequency of stock out occasions and the number of items stocked out annually are to be minimized. For determining the Pareto optimal set, multi-objective evolutionary algorithms are used. First, NSGA-II, MOGA, VEGA, RWGA are developed. Then some improvements in NSGA-II mechanisms are made and R-NSGA-II is developed. Subsequently, these algorithms are examined for some criteria such as set coverage and spacing, and the best algorithms for each criteria arc presented. The Result shows that R-NSGA-II has good scores for most criteria. Afterwards, Pareto optimal set is ranked using the method of global criteria.
خلاصه ماشینی:
Scientific-Research Quarterly of Industrial Management Studies, Year 8, Number 20, Spring 2011, Pages 81 to 99 Presentation of a Multi-Objective Evolutionary Algorithm for Probabilistic Inventory Systems with Continuous Review Soheila Khoshmandeh{o*o} Farhad Farzad{o**o} Mostafa Zandiyeh{o***o} Abstract {IBVarious inventory control systems attempt to determine the order time and quantity in such a way that the highest level of customer service is provided with the minimum cost.
standard normal cumulative probability function {o8o} Inventory Problem If x[-Q,k] is considered the decision vector of the problem and [C(Q,K), N(Q,K), S(Q,K), Q(Q,K)] is the objective vector, the objective functions and decision variable constraints can be defined as follows: (Refer to the page image) {o(1)- o Unit item cost} {o(2)- o Inventory carrying rate} {o(3)- o Lead time} {o(4)- o Lead time demand} {o(5)- o Safety factor} {o(6)- o Safety stock} {o(7)- o Reorder point} {o(8)- o Probability density function of standard normal distribution} Relation 1 relates to minimizing the expected total annual cost, which is obtained from the sum of annual ordering and holding costs.
(refer to page image) Comparison of Algorithm Performance Convergence to Pareto optimal solutions and providing density and diversity among the set of obtained solutions are the two main goals of every multi-objective evolutionary algorithm.
{o(1)- hcaorppa gnimmargorp esimorpmoC o} (refer to page image) Analysis of Results Due to the conflict between the two multi-objective optimization goals, namely approaching the true Pareto optimal set and maintaining the spread and diversity of solutions, an algorithm cannot simultaneously achieve all efficiency criteria.
In addition to the problem model, its solution method can also be expanded; for example, by using other multi-objective evolutionary algorithms and comparing them with the results of the algorithms presented in this article.
(
, 
, 
)
