differential evolution algorithm

The electricity sector plays a crucial role in the country's economy. Therefore, any disruptions in the supply chain of this industry can result in the loss of economic benefits and decrease the competitiveness of industries dependent on this sector. The industry structure analysis of the electricity sector has shown that the lack of proper relationships between entities involved in the supply of goods and equipment leads to disruptions in the electricity supply chain. On the other hand, the specific political and economic conditions of the country, the presence of natural disasters, and high levels of change in the Middle East region have had significant impacts on increasing uncertainty at various levels of the supply chain in this sector. Considering the high uncertainty in the procurement of components in this industry, this study focuses on presenting a set of scenarios for replenishing goods in the supply chain entities of this sector. To achieve this, a probabilistic four-echelon model consisting of a supplier, distributor, retailer, and customer was presented to minimize total inventory costs and the ratio of unmet customer demand based on the (R, Q) policy. Furthermore, by searching organizational documents, interviewing industry experts, and utilizing warehouse management software, data for the model was collected. Subsequently, through experimental design, initial solutions were provided for the differential evolutionary algorithm, and based on this algorithm, different values for reorder points and order quantities were determined. By employing simulation methods, the model's objective values were estimated, and the set of solutions was illustrated in a Pareto chart. Research findings have shown that increasing the average inventory levels of retailers' warehouses leads to a decrease in the proportion of unmet customer demand. This occurs when different reorder point values for two retailers have high levels, but considering the probabilistic demand function, the order quantity can vary. On the other hand, reducing ordering and inventory costs leads to an increase in unmet customer demand. In other words, when reorder point values and low order quantities lead to inventory reduction, customer dissatisfaction increases. The Differential Evolution Algorithm used in this study has accelerated the process of finding solutions and improved model efficiency. This algorithm considers values between high and low levels of reorder points and order quantities, presenting multiple objective function values. Utilizing simulation methods to estimate the probabilistic objective functions employed in the model has increased the speed of executing multiple scenarios, aiding in cost reduction and model execution time. Based on the results of this research, electric equipment with high innovation should have a low reorder point and a high order quantity in the supply chain. This is because the short product lifespan renders the product obsolete along the chain, lacking customer demand and consequently increasing the chain's costs. The computational results of this study indicate that a 105% increase in inventory leads to a 104% increase in customer satisfaction and a 95% decrease in lost sales costs. However, considering the high purchasing costs and large product volume during ordering, a scenario aligned with financial conditions and warehouse capacity should be selected for chain entities.

In the real world, because of decreasing the related cost, the vehicles should return to the depot after serving the last customer’s location. This paper investigates the problem of the increasing service time by using the stochastic time for each tour such that the total traveling time of the vehicles is limited to a specific limit based on a defined probability.

It is proven that classic models in vehicle routing problems (VRPs) belong to the class of NP-hard ones; thus, due to its complexity using exact methods in large-scale problems, a meta-heuristic based differential evolution (DE) algorithm.

The obtained results indicate the efficiency of the proposed DE algorithm.

Originality/Value:

The total travel time is limited to a definite probability percent, and also other constraints (e.g., capacity and time distribution restrictions) are considered while the total cost of the transportation is minimized.

One of the main missions of financial reporting is a proper disclosure of information in order to meet the needs of financial statement users. This study is aimed to investigate whether the company's disclosure quality can be discovered by Machine learning-based models. In this study, disclosure level rating of listed firms is predicted by Securities and Exchange Organization using the ant colony algorithm and differential evolution model. The sample consists of 171 firms listed in Tehran Stock Exchange during the period from 2010 to 2014. This study uses MATLAB software to predict a firm's disclosure quality. The algorithms fitting results show that the two algorithms have ability to predict earnings management with high accuracy of 95%. In fact, the results indicate that the ant colony algorithm has more ability (error 3.316 percent) of predicting earnings management than the differential evolution algorithm (4. 139 percent error).

Current research addresses finding robust and stable schedule for the flexible job shop problem under machine breakdown. We have used simulation to investigate the effect of machine breakdown. Two-stage of metaheuristic algorithm is developed to generate robust and stable schedule and is integrated with simulation algorithm. Because makespan is primitive objective of every scheduling problem، in the first stage of integrated algorithm، makespan is improved and in the second stage linear combination of stability، robustness and makespan is proposed. In our proposed model we provide condition that scheduler can decide which objective is important than the others، then scheduling scheme can be generated based on this decision. In the second stage، we have investigated four type of coefficient in combination cost function، and our experiment shows that it is possible to achieve high stability and robustness measures without sacrificing much from the makespan level. Genetic and differential evolution algorithms are used in proposed model then a statistical hypothesis test is conducted to compare the performance of them.