Modern Research in Decision Making

Abstract DEA is one of the most applicable methods of performance evaluation. In traditional models of DEA, the unit's internal structure and time effects are not considered. Network DEA models survey internal structure, but they are static and do not calculate dynamic efficiency. On the other hand, dynamic models consider the unit as a black box in each period. So, network and dynamic models are not adequate for the evaluation alone. However, the dynamic network DEA models have the advantages of both dynamic and network models. In this study considering a comprehensive dynamic two-stage structure, a dynamic network DEA was presented in both input-oriented and output-oriented, to calculate the optimum weight of input, output, intermediate, and overtime variables. Then using the calculated weight, the method of calculating dynamic network efficiency in each stage, period, and the whole structure was stated. The suggested model for the condition of variables return to scale was also developed and the efficiency relations were explained. In the suggested models, a unit is overall efficient only if it is efficient in all stages and periods. Finally, for the description and potential performance of the suggested models, the data was used from central branches of Keshavarzi bank and the efficiency results were compared in two conditions of CRS and VRS.

Abstract The petroleum products distribution system is the last part of oil supply chain and link in the oil consumption market. Due to this huge supply chain, market uncertainty, the high-risk and ignition nature of oil products and its vital role in the transportation of cargo and passengers, the resilience of this system is remarkable. In this research, resilience is modeled by object-oriented software (Any-Logic) and with a combination of agent-based modeling and GIS. Four agents were defined, which are gas station, main storage, alternative storage tanks and oil tankers. Identification of the distribution process and real data collection has been done in N.I.O.P.D.C in Qom region. By reviewing the existing literature and the oil industry experts’ opinions, resilience measurement indicators have been developed and the demand pattern and storage flow rate have been modeled with a triangular probability density function and the behavior of agents has been defined in the Java programming language. Three scenarios are defined to investigate the effects of disruptions on the system resilience: reducing the number of trucks, the capacity of the main storage and the time to receive orders. The results show that by decreasing the order time by 4 times, the lost sales have increased more than 9 times. In fact, the third scenario is much more effective on resilience than the other two scenarios. This shows the importance of ordering time relative to inventory of gas station in increasing the resilience of the distribution system.

Abstract Cell production is one of the most important applications of group technology that forms production cells in such a way that each family of parts in a cell is processed by a certain group of machines related to that cell (machine cell). In order to have a suitable and coherent plan for a good and quality production, two important steps in production planning should be taken: sequence of operations and scheduling. Sequencing of operations and scheduling is a decision-making process that takes place by allocating resources over time to execute a set of activities in order to optimize one or more objectives. In fact, the main goal in operation sequence scheduling problems is determining the sequence of tasks and allocating resources to tasks in such a way that one or more objectives are optimized. In the proposed model, we intend to design an integrated model of cell formation with cell arrangement and operation scheduling in the cell production system and assigning operators with the aim of minimizing the time to complete the work in addition to paying attention to the non-overlapping of cells and specifying the position. Regarding machines, let's determine the optimal arrangement of the cells independently in our modeling and specify the optimal placement of the cells together. In this research, after nonlinear mathematical modeling and data collection and measurement, the problem is solved by a genetic algorithm, and finally, the best-proposed layout will be selected among the 3 existing scenarios according to the created cost.

Abstract With the emergence of social media, people share their ideas, experiences, opinions, and intentions on Twitter, Facebook, and other platforms. Analyzing and reviewing these opinions on social media gives organizations useful information about the market, customers, and competitors, the purchase intention of users, which for commercial can be used such as purposes like targeted marketing and advertising depersonalization. This research aims to provide a system based on artificial intelligence to predict the purchase intention of users on Twitter. To conduct the research, 23,000 English tweets of users were collected and their sentiments were analyzed using the vocabulary-based method, yielding acceptable results with an accuracy of 0.81. The results of sentiment analysis showed a positive public sentiment about the iPhone 13. The purchase intention of Twitter users was extracted using a long-short-term memory deep convolution neural network with an accuracy of 0.81, and then a model for predicting purchase intention was presented using sentiment analysis and features extracted from tweets. To measure the effectiveness of users' emotions and static features, a multi-layer perceptron deep neural network model was used, and the results showed an accuracy of 0.80. Using this model to market managers plan properly to identify their customers and reduce marketing costs.

Abstract Due to the special position of the scheduling and lot sizing of flow shop scheduling systems in production centers, these issues have received a lot of attention in recent years. One common assumption in these cases is unlimited buffer capacity between different workstations. In industrial environments, the buffer capacity may be limited due to the physical dimensions of the products and the lack of space. Due to the importance of this issue, in the present study the concurrent scheduling and lot sizing of the flexible flow shop scheduling problem considering buffer constraints between workstations and public buffer is investigated which is multi-period and multi-product and the capacity of the machines is also limited. Also, among the buffers, middle buffers have a higher priority to be filled than the public buffer. For the mentioned problem, a Mixed Integer Non Linear Programing (MINLP) model is presented. The objective of this model is to minimize the production, maintenance and external supply costs. The GAMS software is used to solve the model. Due to the complexity of the model and the NP- hardness of the proposed problem, the Hybrid Discrete Artificial Bee Colony (HDABC) is proposed to solve large-scale problems. In order to evaluate the performance of the proposed algorithm, numerical sample problems in different sizes are solved using this algorithm, the GAMS software as well as the genetic algorithm and the Simulated Annealing Algorithm-based Hopfield Neural Network algorithm algorithm. Computational results demonstrate the effectiveness of the proposed algorithm for the considered problem.

Abstract For survival and competitiveness, petrochemical companies must continuously adapt themselves to environmental uncertainties and be resistant to disruptions and instabilities. In this regard, the main aim of the current research is to design and clarify the business continuity management model using the system dynamics approach in the petrochemical industry. At first, the critical success factors of business continuity were identified using a systematic literature review and qualitative content analysis. Then the system dynamics approach was used in Vensim software to design a business continuity management dynamic model. The results showed that the nine identified factors affect the level of business continuity and their implementation increases the level of business continuity. This research could develop theoretical literature on business continuity management and also create helpful insight for managers of petrochemical companies in particular and other organizations in general for the successful implementation of business continuity management and predicting business continuity behavior.

Abstract Implementation of maintenance programs at the right time and simultaneous management and control of inventory, taking into account changes of technology, as well as the use of new technologies, is an issue that can affect the quality of production, as Be considered a competitive advantage. The purpose of this study is to optimize the Expected Loss Rate in the two concept of "maintenance" and "inventory planning and control" based on time and cost. For this purpose, the optimal policy is proposed according to the identified events based on time and cost, using the Markov Decision Process and the values of probabilities in different states of the system. To determine the effectiveness of time and policies, the concept of Industrial IoT has been used and the problem with the OPNET simulator has been modeled and simulated, and based on the new time values, the optimal values have been calculated. For conducting the research, historical data related to the implementation of maintenance and risk assessment in the gas pipeline network have been used. Based on the change in the average occurrence rate of events, the time of simulation and change in the values of network statistical parameters, Sensitivity analysis and model validation are performed. The results of the study indicate the rate of improvement and the optimal rate of the Expected Loss Rate based on time to" implement maintenance policy ", " effect of maintenance policy " and " order spare parts and logistics spare parts", presents based on cost.