Industrial management
Mohsen Kochaki; Behnam Vahdani
Abstract
The correct storage and arrangement of products in the warehouse increase efficiency in responding to requests, accelerate the identification of products, increase accessibility of items in the warehouse, make more use of available space in the warehouse, reduce the possibility of product damage, and ...
Read More
The correct storage and arrangement of products in the warehouse increase efficiency in responding to requests, accelerate the identification of products, increase accessibility of items in the warehouse, make more use of available space in the warehouse, reduce the possibility of product damage, and increase flexibility. The review of studies in the field of warehousing and arrangement of products in warehouses revealed that the use of machine learning algorithms in this field is one of the important research gaps. Therefore, in this article, using machine learning algorithms, we aim to present an innovative algorithm for allocating goods to different parts of a warehouse, for which a real case study is used. The goal of categorizing products based on their characteristics is to minimize the total cost of the system. Hence, spatial clustering algorithms based on the density of applications with noise (DBSCAN), self-organizing mapping neural network (SOM), and AGNES are used. The obtained results show that SOM has better performance than DBSCAN. Also, the DBSCAN algorithm performs better than AGNES.IntroductionWarehouses play a crucial role in every supply chain that involves activities such as receiving, storing, picking, and transporting goods. The way goods are stored directly affects the costs associated with warehousing, so it is important to have efficient management systems in place in order to stay competitive in the global market. Having an organized warehouse layout, utilizing technology for inventory management, and implementing streamlined processes can all contribute to reducing costs and increasing efficiency in warehousing operations. By continuously optimizing operations and staying up-to-date with industry trends, businesses can ensure they are meeting customer demands and staying ahead of the competition (Jinxiang Gu et al., 2007). Storage is the primary and essential function in all warehouses. The methods used for storing items can vary depending on the type of warehouse and its specific goals and objectives (Berman, 1996). The main goal of storage and warehouses is to meet the needs of consumers or enhance service in a manner that takes into account limitations in resources. Efficient management of storage also helps to enhance the speed and reliability of deliveries, which has been identified as a crucial factor for performance in the last twenty years (Ann E. Gary et al., 1992). When looking at logistics costs from an economic perspective, the costs associated with storage and warehousing services make up around 15% of the total logistics costs in developed countries like Germany (Handfield et al., 2013). In this context, properly allocating storage can reduce costs. After deciding how to store the goods, we determine their arrangement. The purpose of this article is to determine the optimal arrangement of goods in the dedicated storage system. Arranging the goods logically in the warehouse increases efficiency in responding to requests, accelerates goods identification, increases accessibility, makes better use of space, determines the location of goods and protects them. It also provides more flexibility and more suitable conditions for storage. It should be noted that due to the functional nature of warehouses, which requires rapid response to determine optimal goods placement, innovative solutions are imperative. All algorithms proposed to solve organizing goods in warehouses must completely consider inclusiveness according to attributes like grouping, similarity, flammability, degradability, inbound/outbound amounts, and stockroom area. Therefore, according to the huge volume and diversity of data in these systems, utilizing data extraction strategies can maximize efficiency of mathematical planning models whose inputs include inbound/outbound amounts for each good and stockroom area assigned. This confirms arrangements account for qualities like item classes, quantities, traits, and warehouse restrictions. Usually, algorithms presented by these methods typically have some limitations. For example, you could reference the inventory of products stocked in your warehouse. A useful way to enhance or address existing issues is through the use of data-driven and machine learning techniques. In this work, we aim to improve an innovative algorithm described in prior studies using data-focused and collaborative learning approaches. Next, we will provide a brief overview of the framework. Then, the problem definition and mathematical model are described. Following, the methods and analyses employed and findings obtained are examined. After, the effect of the algorithm on performance metrics is assessed. Later, applications of machine learning methods for inventory are explained. Finally, results and recommendations are presented.MethodAccording to the items found in the storage facility, nine characteristics for goods were identified, such as group one, group two, similarity, combustion, combustible, corruption, violation, the quantity of goods entering and leaving the warehouse, and storage space extracted. Subsequently, 17 warehouse performance indicators were used to calculate the cost function through a mathematical programming model, analyzing 55 different scenarios. The commodities were then classified using machine learning algorithms SOM, DBSCAN, and AGNES, based on the identified characteristics and inventory performance indicators, with the cost function calculated for each algorithm. Finally, a comparison was conducted between inventory performance indicators and the cost function using the mathematical planning model and the suggested algorithm, with performance evaluated through statistical tests like the Levene test, Kruskal Wallis test, and the Brown for Syte test.ResultsBased on the inventory of 2800 different types of products in the warehouse of Farasan Industrial and Manufacturing Plant, characteristics were extracted for each product. Additionally, warehouse performance indicators and cost functions were analyzed using mathematical programming models and machine learning algorithms. The performance of three algorithms was compared with a mathematical algorithm through statistical tests such as Levene's test, Kruskal-Wallis test, and Brown-Forsythe test. The results showed that the SOM neural network was more efficient than the other two algorithms. Thus, by combining mathematical programming models and machine learning algorithms, one can improve warehouse performance and reduce costs, providing optimal solutions for factory inventory management.ConclusionIn previous research, it was found that products were stored in warehouses without any prior processing. This created a gap in the field, highlighting the importance of categorizing similar goods before storing them in warehouses to reduce storage costs for factories and manufacturing companies. To address this issue, a sophisticated algorithm was developed to enhance product quality in warehouses across all industries. Reducing storage costs is a common objective for companies and factories, influenced by various factors in their environments. This research focused on developing a model for keeping products in warehouses by considering factors such as product diversity. This study used DBSCAN, AGNES, and SOM algorithms to classify products based on 9 features extracted from the products, which resulted in 55 different classification modes with each of the machine learning algorithms. The development of this algorithm aimed to provide factory and warehouse managers with a solution for making more effective decisions in arranging warehouse products.
Amir Daneshvar; Mostafa Zandieh; Jamshid Nazemi
Volume 13, Issue 39 , January 2016, , Pages 1-34
Abstract
Outranking based models as one of the most important multicriteria decision methods need the definition of large amount of preferential information called “parameters” from decision maker. Because of the multiplicity of parameters, their confusing interpretation in problem context and the ...
Read More
Outranking based models as one of the most important multicriteria decision methods need the definition of large amount of preferential information called “parameters” from decision maker. Because of the multiplicity of parameters, their confusing interpretation in problem context and the imprecise nature of data, Obtaining all these parameters simultaneously specially in large scale realistic credit problems which requires real time decision making is very complex and time-consuming.Preference Disaggregation approach infers these parameters from the holistic judgements provided by decision maker. This approach within multicriteria decision methods is equivalent to machine learning in artificial intelligence discipline.Under this approach this paper proposes a new learning method in which Genetic Algorithm(GA) in an evolutionary process induces all , ELECTRE TRI model parameters from training set then at the end of this process, classification is done on testing set by inferred parameters. Experimental analysis on credit data shows high quality and competitive results compared with some standard classification methods.
Jamshid Salehi Sadaghiani; Seyed Amir Reza Abtahi
Volume 4, Issue 13 , June 2006, , Pages 89-122
Abstract
The purpose of this article is about soft computing and its different methods for modeling phenomena. Soft Computing refers to the evolving collection of methodologies used to build intelligent systems exhibiting human-like reasoning and capable of tackling uncertainty.
In this paper, we describe the ...
Read More
The purpose of this article is about soft computing and its different methods for modeling phenomena. Soft Computing refers to the evolving collection of methodologies used to build intelligent systems exhibiting human-like reasoning and capable of tackling uncertainty.
In this paper, we describe the neural networks approach in soft computing at first. Then, other approaches such as genetic algorithm and machine learning will be described. Since the main goal of building the model is knowledge extraction, finally, we will describe the various methods of knowledge and rule extraction from neural networks.