خلاصة:
Location of warehouses and routing of vehicles are two essential key points in order to distribute perishable products properly. Poor location-routing tasks may cause tremendous losses. In this paper, a bi-objective mixed integer mathematical programming is proposed to reduce the total cost of the supply chain and to balance the workload of distribution centers, concurrently. A multi-objective evolutionary algorithm called Non-Dominated Genetic Algorithm-II (NSGA-II) is customized to generate set of non-dominated solutions on Pareto front of the problem. The performance of proposed algorithm and an efficient exact multi-objective method, called ε-constraint, is compared on several benchmark instances using several performance measurements. The analysis reveals the efficacy and applicability of proposed method.
ملخص الجهاز:
"A multi-objective evolutionary algorithm called Non-Dominated Genetic Algorithm-II (NSGA-II) is customized to generate set of non-dominated solutions on Pareto front of the problem.
In this paper, a multi- objective mathematical programming is proposed in order model a location-routing problem for distribution of perishable products.
4 The results of Comparison As mentioned, ten small and medium size test problems presented in Table 2, are solved by proposed NSGA-II and epsilon-constraint method.
Table 3: Results of numerical solution for small and Figure 8: Re-generated Pareto Front on Small and Medium Instances Based on these results, the indices in this study show that the proposed NSGA-II algorithm is efficient and reliable enough to handle large size test and real world problem which are not solvable using epsilon-constraint.
Figure 9 presents the generated Pareto front for a large size instance including 80 customers, 11 potential distribution center, and 14 transportation vehicles using proposed NSGA-II method.
e. an exact method called, epsilon- constraint and an evolutionary computation, called NSGA-II algorithm, were proposed to generate non-dominated solution on Pareto front of instances of the problem.
Other objective functions, such as balancing the number of customers which are served by each distribution center, balancing the time that transportation vehicles are operating, can be modeled in future studies.
In this study, we propose a methodology based on a neural network, a modeling technique, and four different genetic algorithms for multiple response statistical optimization problems when the problem have stochastic nature and the relationship between control factors and the responses is unknown and so complicated."