Journal of Operational Research in its Applications
Year:2019 | Volume:16 | Issue:2 (61) |Papers Count:8

The efficient planning of each system requires an understanding of the true nature of the system. Regression models also represent a system that its input variables, the role of the system input, its variables coefficients, the role of the system process, and its response variables, play the role of the system output. Therefore, for exact regression modeling and efficient planning, it's essential that the components of the model, including variables, target circles, and constraints, are correctly identified and defined. One method of the regression modeling is the design of experiments and the response surface method that are widely used to identify complex systems. In the present research, we intend to introduce a guidelines to improve the determination of the experiment range of control variables and measure the effect of experiment range of control variables to efficiency of regression modeling. The results showed the positive role of introduced guidelines and high effect of range of control variables to efficiency of regression modeling in the form of numerical results that are an approach for experts of mathematical and regression modeling, design of experiments, and response surface methodology.

In this paper, mathematical models are proposed for simultaneously modeling location and inventory control decisions in a four echelon supply chain network considering capacity. The echelons considered in this paper include suppliers, warehouses, hubs and retailers. The aim of the model is to minimize the location, transportation and inventory control costs. Hence, a non-linear mixed integer programming model is provided. The proposed model is then transformed to a conic mixed integer programming model in order to benefit from the higher solving speed of the commercial packages. To compare the different conditions of the inventory control, lost sale and delay strategies are also considered. The computational results show the reliability of the model in different inventory strategies and shorter required solution time for conic quadratic model. Therefore, the application of this formulating can be beneficial for large scale problems using commercial software.

In this paper, a constrained shortest path problem (CSP) in a network is investigated, in which some special plans for each link with corresponding pre-determined costs as well as reduction values in the link travel time are considered. The purpose is to find a path and selecting the best plans on its links, to improve the travel time as most as possible, while the costs of conducting plans do not exceed the available budget. Using the Lagrangian relaxation approach, some constraints of the problem are relaxed and the Lagrangian dual problem is decomposed into two smaller sub-problems. Then, by applying the sub-gradient algorithm, a near optimal solution is determined for the original problem. Finally, by considering the proposed model on a small-sized network and on Khorasan state network, solutions for different origin-destination pairs with different parameters are determined.

It is very necessary to consider the uncertainty in the data and how to deal with it when performance measurement using data envelopment analysis. Because a little deviation in the data can lead to a significant change in the performance results. However, in the real world and in many cases, the data is uncertain. Interval data envelopment analysis is one of the most widely used approaches to deal with interval data uncertainty. The purpose of the present paper is to provide a robust model of interval data envelopment analysis in order to performance measurement under double uncertainty situations. In addition to the uncertainty caused by the interval of data, there is also uncertainty at the lower bound and upper bound of the interval for each data. Using the approach presented in this study can greatly increase the conservatism and validity of the efficiency results and ranking. Finally, it should be noted that the results of the proposed models are illustrated by using of a numerical example.

In social networks and big size organizations, finding the number of members that all of them have a certain relationship is an important problem for managers, as well the number of members that none of them has not the relationship. Considering the members as vertices and the relationship as edges, and denote the clique number and the number of edges in the independent graph, respectively. and can be find by solving some optimization models. One of these models is copositive programming. Unfortunately, there is no any polynomial time algorithm for solving these problems. In this paper, we model these problems as a copositive programming then solve them using genetic and particle swarm algorithms. Illustrative examples show the efficiency of our method.

Trust region methods are a class of important and efficient methods for solving unconstrained optimization problems. The efficiency of these methods strongly depends on the initial parameter, especially radius adjusting parameters. In this paper, we propose a new strategy for choosing the radius adjusting parameters. Numerical results from testing the new idea to solve a class of unconstrained optimization problems from CUTEst package [7] show that the new idea increases the efficiency of trust region methods.

Risk analysis is necessary for portfolio management. Investors in setting an appropriate portfolio need to pay attention to risk-related indicators as well as profitability. Beside soft approaches for composing indicators to an index such as AHP, the optimization approaches such as DEA are also considered. However, if the indicators have a hierarchical structure, then it is no longer possible to use the basic DEA model for weighting and ranking. Therefore, it is necessary to extend the DEA model to be able to consider the hierarchical structure of indicators and sub-indicators. In this paper, by reviewing the various risk indicators in the banking sector such as credit risk, liquidity risk, and profitability risk and also by using the multilayer DEA (MLDEA) approach, a Composite Risk Index (CRI) will be constructed. To indicate the capabilities of the MLDEA model, 10 Iranian banks in Tehran Stock Exchange were evaluated. The results indicated the Pasargad has the best performance (lowest Risk) among all other banks under evaluation. In this bank, Liquidity risk is the best, and credit risk and profitability risk are second and third, respectively. In addition to providing the CRI score and the weights of sub-indicators, the discriminative power of the model has also been significantly improved.

In this study, first by using the finite element method, fatigue life of fiber metal laminates of Glare type subjected to impact was obtained and the numerical results of the model were compared with the experimental results. With regard to the very good matching between numerical and experimental results, then the results of the finite element model were generalized and expanded, and with using of the multilayer neural network, numerical model was extracted and then by applying the meta-innovative algorithm, the maximum fatigue life of glare was determined at highest level of very low velocity impact.