Abstract: The location decision of logistics distribution center plays an important role in logistics operation. In this paper, the literature about the location selection method of distribution center at home and abroad in recent years is sorted out and studied. The results show that each site selection method has its own advantages and disadvantages and a certain scope of application, and the combination of various methods is the future research trend in this field.
Keywords: Literature Review of Logistics Distribution Center Location
In the operation of logistics system, the location decision of distribution center plays an important role. The distribution center is the middle bridge connecting factories and customers, and its location often determines the distribution distance and mode of logistics, which in turn affects the operation efficiency of logistics system. Therefore, it is of great theoretical and practical significance to study the location of logistics distribution center.
In this paper, the domestic and foreign literatures about the location selection methods of logistics distribution centers in recent years are sorted out and studied, and various methods are compared. There are two main methods for site selection: qualitative and quantitative. Qualitative methods include expert scoring method and Delphi method. Quantitative methods include barycenter method, P- median method, mathematical programming method, multi-criteria decision-making method, various heuristic algorithms for solving NP-hard problems (uncertainty problems of polynomial complexity), simulation method and their combinations. Because the qualitative research methods of barycenter method and P- median method are mature, this paper will mainly analyze the research status of mathematical programming, multi-criteria decision-making, various heuristic algorithms for solving NP-hard problems and the application of simulation in distribution center location.
Mathematical programming method
Mathematical programming algorithms include linear programming, nonlinear programming, integer programming, mixed integer programming and dynamic programming, network programming algorithms and so on. In recent years, the concept of uncertainty is often introduced into planning theory, which further produces fuzzy programming, stochastic programming, fuzzy stochastic programming, stochastic fuzzy programming and so on. Uncertainty planning mainly introduces uncertainty into C (value vector), A (resource consumption vector), B (resource constraint vector) and decision variables in planning, thus making uncertainty planning closer to the actual situation and being widely used in practice.
Scholars at home and abroad have conducted in-depth research on the application of mathematical programming methods in the location of distribution centers. Jiang Dayuan (2005) applied baumol-Wolff model to study the location problem of multi-logistics nodes, and illustrated the application of the model through an example. This model is a combination of integer programming and non-parametric programming. In the practical application of various planning methods, NP-hard problems often appear. Therefore, the further research trend at present is to comprehensively plan and calculate the location of distribution centers by combining various planning methods and heuristic algorithms.
Multi-criteria decision method
In the research of logistics system, people often encounter a lot of multi-criteria decision-making problems, such as the location of distribution center, the choice of transportation mode and route, the choice of suppliers and so on. The typical feature of these problems is that they involve multiple options (objects), and each option has several different standards. We should make a comprehensive choice of options (objects) through multiple criteria. For the location of logistics distribution center, people often make decisions based on the criteria of minimizing the sum of transportation cost, construction and operation cost of distribution center, meeting customer needs and meeting social and environmental requirements. The methods of multi-criteria decision-making include multi-index decision-making method and multi-attribute decision-making method, and the commonly used methods include analytic hierarchy process, fuzzy comprehensive evaluation method, data envelopment analysis method, TOPSIS method and priority method.
Multi-criteria decision-making provides a good decision-making method system, which has been widely used and studied in the practice and theoretical research of distribution center location. Guan Zhimin et al. (2005) put forward the optimization decision of distribution center location based on fuzzy multi-index evaluation method. Based on the actual needs of supply chain management, this paper analyzes the main factors affecting the location of distribution centers, and establishes the corresponding evaluation index system, thus giving a fuzzy multi-index evaluation method combining qualitative and quantitative. Chen-Tung Chen(200 1) used fuzzy multi-criteria decision-making based on triangular fuzzy numbers to study the location problem of logistics distribution centers. In this paper, the investment cost, the possibility of expansion, the convenience of obtaining raw materials, and the proximity between human resources and customer market are taken as decision-making criteria, and the weights of each criterion are aggregated by semantic fuzzy judgment.
Multi-criteria decision-making methods, especially analytic hierarchy process and fuzzy comprehensive evaluation method, are widely used in the study of distribution center location. However, these two methods are based on the idea of linear decision. In today's complex and changeable environment, linear decision-making thought gradually exposes its inherent limitations, and nonlinear decision-making method is the focus and trend of further research in the future.
Heuristic algorithm
Heuristic algorithm is a method and strategy to solve problems based on experience and judgment, which embodies people's subjective initiative and creativity. Heuristic algorithms can often deal with NP-hard problems effectively, so heuristic algorithms are often combined with other optimization algorithms to further develop their advantages. At present, the commonly used heuristic algorithms include: genetic algorithm; Neural network algorithm; Simulated annealing algorithm.
(A) Genetic algorithm
Genetic algorithm (GA) was put forward in 1960s. It is a search algorithm inspired by natural selection and genetic mechanism in genetics. Its basic idea is to solve complex optimization problems by simulating the evolution of organisms and humans, so it is also called simulated evolutionary optimization algorithm. Genetic algorithm mainly has three operators: selection; Cross; Variation. Through these three operators, the problem is optimized step by step, and finally a satisfactory optimization solution is obtained.
Many scholars at home and abroad combine genetic algorithm with general planning method to study the location of logistics distribution center. Jiang Zhongzhong (2005) established a mathematical programming model (mixed integer programming or general linear programming) on the basis of considering various costs (including transportation costs). ) and combined with the specific application background. Because the model is a combinatorial optimization problem of NP-hard problem, genetic algorithm is combined to solve the model. By choosing appropriate coding methods and genetic operators, the optimal solution of the model is obtained.
As a random search and heuristic algorithm, genetic algorithm has strong global search ability, but it is often easy to fall into local optimum. Therefore, in research and application, in order to avoid this shortcoming, genetic algorithm is often combined with other algorithms, making this algorithm more valuable.
(2) Artificial neural network
Artificial Neural Network (ANN) is a network which is widely interconnected by a large number of processing units (neurons). It is an abstraction, simplification and simulation of the human brain, reflecting its basic characteristics. By learning the sample training data, a certain network parameter structure can be formed, so as to effectively identify complex systems. After a large number of samples learning and training, neural networks are often more effective than general classification and evaluation methods.
Many scholars at home and abroad have made various useful attempts on how to apply neural network to the location selection of logistics distribution centers. Han Qinglan and others (2004) tried to study the location problem of logistics distribution center with BP network, which showed that it was feasible and operable to solve the location problem of distribution center with neural network.
The deficiency of this study is that the training of neural network needs a lot of data, and it is not suitable to use neural network for learning when the data is difficult to obtain. When applying ANN, we should pay attention to the learning speed of the network, whether it is trapped in the local optimal solution, the data preparation in the early stage, the structural explanation of the network, etc., so as to effectively and reliably apply ANN to solve practical problems.
(3) Simulated annealing algorithm
Simulated Annealing (SA), also known as simulated cooling method and probability climbing method, is another heuristic stochastic optimization algorithm proposed by Kirpatrick in 1982. The basic idea of simulated annealing algorithm is the process of starting from an initial solution, repeatedly generating iterative solutions, gradually judging and abandoning, and finally obtaining a satisfactory solution. Simulated annealing algorithm can develop in both good and bad directions, thus making the algorithm jump out of the local optimal solution and reach the global optimal solution.
For the research on the application of simulated annealing algorithm to the location selection of logistics distribution center, a large number of documents are combined with other methods (such as multi-criteria decision-making, mathematical planning, etc.). ) has been studied. Ren Chunyu (2006) proposed a method of combining quantitative simulated annealing genetic algorithm with analytic hierarchy process to determine the address of distribution center. This method ensures the diversity of individuals in the population and prevents the premature convergence of genetic algorithm. The weight of location evaluation index of logistics distribution center is determined by analytic hierarchy process, and comprehensive evaluation is made by combining with expert scoring. The algorithm is effective and reliable for solving the location problem of logistics distribution center.
In addition to the above three commonly used methods, heuristic algorithms also include ant colony algorithm, tabu search algorithm, evolutionary algorithm and so on. There are some differences in global search ability, advantages and disadvantages, parameters and solving methods of various algorithms. All kinds of heuristic algorithms basically have the characteristics of random search, which are widely used to solve NP-hard problems, and also provide the possibility for intelligent processing of logistics distribution center location. It is a feasible and operable research method to establish mathematical model by analytical method (including linear programming) and then solve it by heuristic algorithm.
Simulation method
Simulation is to use a computer to run a simulation model to simulate the running state of a time system and its changing process with time. Through the observation and statistics of the simulation running process, the simulation output parameters and basic characteristics of the simulated system are obtained, so as to estimate and infer the real parameters and real performance of the actual system. Many literatures at home and abroad have applied the simulation method to the research on the location of logistics distribution centers or general facilities, and the research results are closer to the actual situation than the analytical method.
Zhang Yunfeng et al. (2005) studied the location of distribution center of automobile group enterprises by simulation method. Firstly, several schemes of distribution center location are determined, and the simulation models of each scheme are established by using Flexim software. According to the simulation results, the scheme is analyzed and selected. This method provides an ideal solution for the location problem of distribution center in group enterprises. Xue Yongji et al. (2005) studied the optimal number of stations in the logistics center by establishing a mathematical model, and solved the optimal number of stations under certain assumptions and a series of restrictions. In view of the complexity of mathematical model and the shortage of solving methods, a simulation model is established on ARENA simulation software platform to determine the optimal scheme. Kazuyoshi Hidaka et al. (97) used simulation to study the location of large warehouses. In this study, the fixed cost and transportation cost of the warehouse are simulated and 6800 customers are satisfied at the same time to find a near-optimal solution. In the process of solving, the greedy exchange heuristic algorithm and balloon search heuristic algorithm are combined to solve the problem. The algorithm can effectively avoid falling into the local optimal solution and get a satisfactory location scheme. However, the research results are easily influenced by the change of average speed of transport vehicles.
Compared with analytical method, simulation method has some advantages in practical application, but it also has some limitations. For example, the simulation needs a relatively rigorous test of the credibility and effectiveness of the model. Some simulation systems are sensitive to the initial deviation, which often makes the simulation results differ greatly from the actual results. At the same time, the requirements of simulation for people and machines are often relatively high, which requires designers to have rich experience and high analytical ability, and relatively complex simulation systems also require high computer hardware. For future research, the combination of various analysis methods, heuristic algorithms, multi-criteria decision-making methods and simulation methods is an inevitable trend. The combination of various methods can make up for their respective shortcomings and give full play to their respective advantages, thus improving the accuracy and reliability of site selection.
The location decision of logistics distribution center has an important influence on the operation of the whole logistics system and customer satisfaction. On the basis of studying the domestic and foreign literatures on the location methods of logistics distribution centers, this paper compares and analyzes the applications of mathematical programming method, multi-criteria decision-making method, heuristic algorithm and simulation method in the location selection of distribution centers. It is found that mathematical programming method, multi-attribute decision-making method, heuristic algorithm and simulation method have their own advantages and disadvantages and a certain scope of application, and the combination of these methods is a trend in future research. At the same time, due to the dynamic, complex and uncertain characteristics of the location problem itself, developing and studying new models and methods is also a necessary way to further solve the location problem of distribution centers.
References:
1. Jiang Zhongzhong, Wang, Optimization model and algorithm for location selection of B2C e-commerce distribution center (J). Control and decision-making, 2005
2. Han Qinglan, Mei Yunxian. Location Decision of Logistics Distribution Center Based on BP Artificial Neural Network (J). China Soft Science, 2004.