Optimal international logistics service composition algorithm based on improved particle swarm optimization algorithm in cloud environment

Abstract

Under the environment of cloud, particle swarm algorithm is widely used in intelligent computer field. The combination model of the logistics service is solved. However, in solving workflow and call problems, the traditional algorithm consumes more time and does not meet the logistics application scenarios. In this paper, the particle swarm optimization algorithm was optimized and improved. The execution order of users was used to re arrange the algorithm. The experimental results showed the high efficiency of the algorithm and rapid computing speed. In order to solve the problem of particle swarm algorithm “premature”, a jamming algorithm was designed in this research. When the similarity of particle swarm was greater than a limit value, the particle position was updated optimally, and the local optimal solution and global optimal solution were retained. The particle swarm optimization algorithm could successfully avoid that the particle swarm optimization got into the local dead loop problem when searching for the optimal solution. It could be seen based on particle swarm optimization algorithm experimental results that the algorithm had high superiority in computational efficiency and speed.

Keywords

Particle swarm optimization service composition international logistics user preference

1 Introduction

At present, the development of the city can’t be separated from the improvement and development of the logistics system. At the same time, logistics plays a very important role in the development of the whole society and economy [1]. Logistics industry is also the basic pillar of the development of various industries. Logistics industry involves a very wide range, and the crowd involved is also more complex, which can promote the development of the third industry, driven the crowd consumption and promote production. The development of logistics industry is closely related to other sectors and branches in social production. These structures compete with each other and depend on each other, and the relationship is complicated. And logistics is the bridge of various industries [2]. The development of the national economy requires continuous consumption and stimulating consumption. The process of obtaining the means of production by enterprises is the transit process of the logistics. Logistics can be implemented smoothly in this process. The development of market needs the circulation of commodities, so that it can adapt to the corresponding benign operation. Logistics development is of vital importance. The carrying capacity of railways and highways reflects the quantity of steel and coal production in an area [3]. The rationalization of logistics arrangements can promote the circulation of production materials, make quantitative and timely supply according to customer demand, and ensure the smooth production and operation of various industries.

Logistics costs of Japan and western countries in the industry operation can be controlled at around 8%, while China’s logistics cost control is about 20%. In such a high cost logistics consumption situation, if it can’t be effectively controlled and improved, China’s enterprises will encounter unprecedented operational pressure in the future [4]. Slowly, the labor cost advantage of China’s low cost labor force will be slowly offset, which seriously restricts the speed of transnational development of China’s enterprises. The bottleneck faced by the world development is also extensive logistics management model. In this paper, the research was carried out on logistics combination and algorithm. Through the optimization and improvement of the algorithm, the logistics operational efficiency can be promoted and the logistics application is more reasonable. The promotion of enterprise supply chain collaborative logistics development and management can be obtained. The development of enterprise collaborative logistics was promoted, which is very beneficial to the logistics management and development of small and medium-sized enterprises, and it is of great significance for the development and management of modern logistics with high flexibility [5]. Reasonable logistics management and use can reduce the inventory of enterprises and reduce the amount of loss of goods. Speeding up the turnover of goods can improve the quality of goods and achieve the minimum storage capacity of goods [6]. This study analyzed the characteristics of logistics services in cloud environment. After the particle swarm optimization is improved, how to quickly provide logistics services to meet customer needs is solved.

2 Technology analysis

2.1 Cloud flow service

The characteristics of modern logistics are obvious. Compared with the traditional logistics, modern logistics can be closely integrated with e-commerce, and adapt to the needs of social progress and economic development. Modern logistics is a combination of automation, networking and intelligence. It has the characteristics of fast response, integration and modernization and obvious characteristics [7]. The storage and transportation of traditional logistics activity patterns consider the value orientation of the enterprise as the basis for development. The development of modern logistics considers information technology as the mainstream, and makes emphasis on information integration operation. Modern logistics will be organically integrated with social logistics and brings the meaning of logistics to new meaning, and makes cross departmental collaboration to form a new social system [8]. Starting from procurement, through the production, sales, logistics and then through transportation and warehousing process, a consumption process is formed. Under the premise of meeting the needs of users, through reasonable planning and overall control of commodity circulation process, it is necessary maximize the interests and minimize the cost control. The development trend of modern international logistics requires the systematization of logistics as a whole, which is beneficial to the promotion of circulation efficiency. Logistics is developing toward information and globalization, and the transformation speed of production factors and commodities is increasing [9].

Traditional logistics management mode is only concerned about its own resource status and manages procurement and production and sales departments and makes simple planning of enterprise internal resource operation and management [10]. Logistics management of the internationalization is the whole supply chain. The global logistics management is impossible to manage each department or node details, so the supply chain management is more suitable for international logistics resource management. According to the traditional mode of logistics development, the enterprises run independently and compete with each other. In the modern international logistics development strategy, the cooperation and mutual aid of the node enterprises in the supply chain are advocated [11]. The internationalization development strategy relationship of enterprises is built to achieve win-win situation and the interests of technology development. The international logistics service mode is not only the cooperation between enterprises and enterprises, but the combination of each node [12]. International logistics is an organic whole of supply chain. It is not a simple superposition, but an integrated approach is used to realize the full application of the whole supply chain.

The supply chain is able to take advantage of the overall management advantage, and integrates inventory and distribution of cross border enterprise orders into a detailed process from procurement to manufacturing. The management model of the supply chain is shown in Fig. 1. The division of labor among enterprises is detailed. Technical standards will be reduced to make labor intensive jobs entre into low-cost developing countries [13]. It hopes the parts can be supplied on time with good quantity. That task is beyond the management ability of a complete enterprise. It requires close cooperation among enterprises and ensures smooth logistics sharing.

Fig. 1

Supply chain management model.

In this paper, it hoped to complete the overall planning of the international logistics management program in the cloud environment, which was shown in Fig. 2. The service process provides the planning and utilization of equipment and human resources. In the software resources and other aspects, the WSDL service system is used to release their own supply and demand. The service demand represents user requirements on logistics node. The user puts forward the index demand fitting preference of logistics to the Internet cloud service system, and WSDL lookup service obtains results from the system and makes feedback to service personnel [14]. According to the demand relationship agreement established by the service providers, the system evaluates and manages the logistics services of each node. In particular, the calculation of the combination of logistics services needs to be implemented accurately in order to ensure the quality of logistics services.

Fig. 2

Logistics service overall planning model.

In order to understand the Web process of the cloud environment more clearly, Fig. 3 gives the flow chart of the order management in the logistics service. Each Web can realize an independent unit. The task to show the public interface is provided by logistics service. For practical cooperation close business enterprise, logistics service is provided by service provider [15]. Static binding method is used and it is necessary only to simply locate the cooperative enterprise, and then the WSDL service is quoted. WSDL service transforms a role of a service provider. The static binding of WEDL services can provide service interface in the stage [16].

Fig. 3

Logistics order execution process.

2.2 Cloud logistics service index weight setting

Logistics cloud service technology features is able to run a dynamic selection method. Web service model of combinatorial logistics is a Qos mechanism that can be combined with functional requirements. Dynamic selection of logistics cloud services is carried out. Although the specific application domain has the Qos standard, the model needs a certain extension. After adding the Qos model standard, it can’t change basic service standard [17]. In order to confirm the feasibility of dynamic binding of cloud services in logistics, a dynamic choice model for Qos logistics cloud services was proposed in this paper. The combination of logistics services model is given to calculate, and the best logistics cloud service can be obtained from which [18]. Logistics cloud services provide users with an execution price of service operations, the basic costs required to execute and complete certain tasks. The term expression of logistics execution is as follows [19]: $q_{du} (s, t) = T_{process} (s, t) + T_{trans} (s, t)$ (1)

In the formula, T_process (s, t) represents the processing time of the technical service, and T_trans (s, t) represents the time of the delivery.

The expression of the transfer time is as follows [20]: $T_{t r a n s} (s, t) = \frac{\sum_{i = 1}^{n} T_{i} (s, t)}{n}$ (2)

The calculation expression for the reliability of logistics service mode is as follows: $q_{rel} (s) = N_{invoke} (s) / k$ (3)

In the formula, N_invoke (s) represents the number of times a service (s) is called (k) within the specified time period.

In this paper, the vector expression of logistics service quality is given as follows: $Q (s) = (q_{price} (s, t), q_{du} (s, t), q_{rel} (s), q_{av} (s))$ (4)

In the formula, q_av (s) represents the availability of logistics services, which can be understood as the basic probability value of logistics being accessed by users.

In the process of logistics service combination, due to the existence of Web of multiple standby services, it is necessary to consider the index criteria of spatial factors, and whether the optimal service can be searched. For stochastic logistics tasks, the quality can be expressed by the following formula: $Q_{t} = {\begin{matrix} Q_{1, 1}, Q_{1, 2}, Q_{1, 3}, Q_{1, 4}, Q_{1, 5} \\ Q_{2, 1}, Q_{2, 2}, Q_{2, 3}, Q_{2, 4}, Q_{2, 5} \\ \dots, \dots, \dots, \dots, . . \\ Q_{n, 1}, Q_{n, 2}, Q_{n, 3}, Q_{n, 4}, Q_{n, 5} \end{matrix}}$ (5)

It can be seen from formula 5 that the Qos model of spare logistics service can be decomposed into a kind of extended matrix. The column is the quality standard Qos value, and the line is the vector value of the model. The logistics service quality standards are divided into two categories and they are the availability and reliability standard of logistics services. The higher the Qos value is, the better the service quality of logistics is. However, in view of the execution time for logistics and logistics price standard, the higher the Qos value, the lower the quality is.

Different types of quality standards and the describe degree are inconsistent. The unified measure of logistics service level is carried out, and the weight calculation expression by using Qos is as follows. $V_{ij} = \frac{Q_{ij}}{| Q_{J}^{max} |} - \frac{1}{2} (\frac{Q_{j}^{max}}{| Q_{j}^{max} |} - 1)$ (6)

In the formula, the absolute maximum value of service quality is $Q_{J}^{max}$ . And V_ij is the weight value corresponding to Q_ij calculation.

In order to obtain the optimal combination service model of service, the comprehensive quality mark of computing service is calculated. The standard weight of each category is W_j. In order to design a basic structure model with a certain degree of preference to users, the service quality of logistics was evaluated in this paper. Integrated logistics service marks metric calculation expression is as follows. $Score (s_{i}) = \sum_{j = 1}^{n} (V_{ij} * W_{j})$ (7)

In the logistics service with the condition of n, the optimal calculation method of network cloud logistics service is as follows. $S_{e x e l e n c e (i)} {= \max}_{i = 1}^{n} S c o r e (i)$ (8)

3 Combination calculation method of logistics service in cloud environment

3.1 Logistics combination algorithm

Firstly, the standard weight model of experts is given before the algorithm is given. The purpose of this design is to solve the problem with lack of professional knowledge to the users. In general, only when the logistics index changes, adjustments will be made. Figure 4 is the standard evaluation matrix model. It can be seen in Fig. 4 that E_i denotes the value of the i index. x_ij denotes the importance weight value of the first i index corresponding to the j index. we_i represents the standard weight of experts after conversion calculation.

Fig. 4

Standard evaluation matrix.

Under the premise of ensuring users’ subjective and individual needs and meeting the characteristics of professional fields, it is necessary to combine the evaluation weights of experts with the evaluation of users effectively. The comprehensive preference weight is SYN, and calculation expression is as follows.

$S Y N = (\frac{w e_{1} \times w s_{1}}{\sum_{i = 1}^{n} w e_{1} \times w s_{i}}, \frac{w e_{2} \times w s_{2}}{\sum_{i = 1}^{n} w e_{i} \times w s_{i}}, \dots, \frac{w e_{n} \times w s_{n}}{\sum_{i = 1}^{n} w e_{i} \times w s_{i}})$ (9)

The logistics reliability, logistics service reputation, execution time and logistics cost in the selection matrix are taken as the QOS indexes. Table 1 is the model table of fuzzy matrix evaluation.

Table 1

The model of blurry and comparative evaluation

QOS	Perform cost	Execution time	Fame	Reliability	Usability
Perform cost	0.5	0.4	0.7	0.6	0.5
Execution time	0.6	0.5	0.4	0.6	0.3
Fame	0.3	0.6	0.5	0.7	0.6
Reliability	0.4	0.4	0.3	0.5	0.5
Usability	0.5	0.7	0.4	0.5	0.5

In the corresponding matrix evaluation, the fuzzy matrix complementary model is obtained by index calculation. The form of the fuzzy matrix is shown in Fig. 5.

Fig. 5

Fuzzy complementary matrix.

The calculation formula of the matrix transformation formula consistent with the fuzziness is as follows. $a_{i} = \sum_{m = 1}^{n} a_{im}, (i = 1, 2, \dots, n)$ (10) $f_{ij} = \frac{a_{i} - a_{j}}{2 (n - 1)} + 0.5$ (11)

After calculation, the fuzzy matrix data table can be obtained:

The calculation method of fuzzy consistent matrix is as follows. $F = (f_{ij})_{n \times n}$ (12)

Through the following calculation method, the expert evaluation index weight is obtained:

$\begin{matrix} w e_{i} = \frac{\sum_{s = 1}^{k} \sum_{i = 1}^{n} \sum_{j = 1}^{n} a_{s} f_{i j} + \frac{n}{2} - 1}{n (n - 1)}, & i = 1, 2, \dots, n \end{matrix}$ (13)

3.2 Particle swarm optimization algorithm

Ant colony algorithm, genetic algorithm and particle swarm algorithm are widely used in the combination algorithm of Web services. In this paper, the Web services of linear programming were selected, and the linear parameters were weighted to be transformed into a single objective function. Linear programming is used to solve the problem of the whole optimization model. The operability of the algorithm itself is relatively strong. The probability of searching the optimal solution by genetic algorithm is relatively large, but it takes a long time. In addition, the complexity is relatively high. Ant colony algorithm is more suitable for local optimal solution and combination method, and has very good practicability. But in the initialization of parameters, many of them are using experience to analyze the problem and lacks of theoretical support and textual research. In order to solve the above problems, an improved particle swarm optimization algorithm is proposed, and a method to resist premature convergence is designed so as to be able to better solve the logistics network service global optimization problem.

Table 2
The grading of qos targets

QOS Perform cost Execution time Fame Reliability Usability

Perform cost 0.5 0.4 0.575 0.45 0.415

Execution time 0.6 0.5 0.375 0.55 0.325

Fame 0.425 0.625 0.5 0.725 0.6

Reliability 0.55 0.45 0.275 0.5 0.535

Usability 0.585 0.675 0.4 0.465 0.5

QOS	Perform cost	Execution time	Fame	Reliability	Usability
Perform cost	0.5	0.4	0.575	0.45	0.415
Execution time	0.6	0.5	0.375	0.55	0.325
Fame	0.425	0.625	0.5	0.725	0.6
Reliability	0.55	0.45	0.275	0.5	0.535
Usability	0.585	0.675	0.4	0.465	0.5

Particle swarm algorithm is a relatively well-known intelligent algorithm. Because it can keep the high calculation speed, and has attracted extensive research in many aspects. The algorithm is based on group iteration. The basic idea of PSO algorithm is to initialize the particle. The particle itself does not have the volume and quality. The individual particles can be considered as a set of feasible solutions to the problem, and the superiority of the particles is calculated according to the fitness function. When the particle activity is carried out in the feasible space, the speed of change determines the distance and direction of particle movement. Particles automatically chase the best particle motion, and then find the most computational solution according to gradually updating and search. Each class of particle has two extremes. One extreme value is the optimal solution found by the particle itself, and the other extreme is the optimal solution found by the group. In ordinary particle swarm optimization (PSO), x_i = (x_i1, x_i2, …, x_in) expresses the specific location of the i particle, which determines the fitness function and estimates the superiority of the particle position. v_i is the flight speed of particles. The best location particle to search is pbest and the best location of the whole population to search is gbest. The standard particle swarm algorithm location update formula is as follows. $\begin{matrix} v_{id}^{k + 1} & = & v_{id}^{k} + c_{1} r_{1} (pbest - x_{id}^{k}) \\ + c_{2} r_{2} (gbest - x_{id}^{k}) \end{matrix}$ (14) $x_{id}^{k + 1} = x_{id}^{k} + v_{id}^{k + 1}$ (15)

In the formula, r₁, r₂ are optional random values in the range of [0, 1], and c₁, c₂ are the learning factors in the formula.

The optimized particle swarm algorithm is used to solve the most optimal service solution to meet the demand in the logistics service set of each node. The solution obtained is the set of the selected services of each node. Since the discrete data is interpreted, it is inappropriate to adopt particle swarm optimization directly. Suppose the number of nodes is n, and {ws_(i)1, ws_(i)2, …, ws_(i)k, …, ws_(i)m } represent logistics service standby selection set of the node i. k represents the k service of an alternate candidate node. The number of instances of logistics services is m.

The l of Qos indexes is evaluated, and the evaluation methods are as follows. $max {Qos}_{l} ({ws}_{o}), 1 \leq l \leq t$ (16) $min {Qos}_{l} ({ws}_{o}), 1 \leq l \leq t$ (17)

max function can be transformed into min function by transformation. The weights of Qos indexes are determined according to the degree of users’ preferences, and the objective optimization problem is converted into a single objective optimization problem. w_l represents the set weighting value of the model, and the expression of fitness function is given here. $f (x) = min \sum_{l = 1}^{t} w_{l} {Qos}^{,} ({ws}_{o})$ (18)

In order to avoid premature maturity in particle search or it is the maximum to avoid premature particle entry into local optimal solution and calculation, a method of interference was designed in this paper. When the similarity degree of particle swarm is greater than some maximum value, the motion of particle is in close state, and the position coordinates of particle need to be re corrected. The local optimal solution and the global optimal solution are retained here, and the optimal service of the company is the calculated letter vector of each particle. The comparative analysis method using similarity of two vectors is used. No letter similarity at the same location is 0, and letter similarity when there is a group with the same position is 1. Letter similarity with two group records is 2. In this way, it continues to infer downward. The similarity calculation method of particle swarm is as follows. $\bar{s} = \sum_{i = 1}^{M} \sum_{j = 1}^{M} S_{ij} / M^{2}$ (19)

The data processing is carried out in advance, and the data which will not meet the basic constraints of are screened out at the first step. Specific services of service reserve set of each node are made number, and a random position is given to the particle. The number of particles is set to n. Local optimal solution before particle searching is pbest, and the global optimal solution is gbest. After updating method, the latest fitness function is used to calculate the particle position. Assuming that the result is greater than pbest, a number of vectors are generated randomly, and then the letters position are generated to generate the latest position. Suppose the result is less than pbest, the results are replaced to generate different positions as the new location of particles. The maximum parameter is determined and the similarity of the particle swarm is calculated. Assuming that the similarity of the particle swarm is greater than the extreme value, the particle position of the particle swarm needs to be updated. Each generation of particles needs to adapt to the global optimal value function calculation. Finally, the global optimal gbest is determined by the minimum value of fitness. Particle swarm optimization algorithm obtains the optimal solution by performing iteration. The specific execution process is shown in Fig. 6.

Fig. 6

Flow chart of optimized particle swarm optimization algorithm.

In view of particle swarm optimization, the data are calculated through experiments. Due to the ranges of logistics time, logistics prices, accurate use of logistics and other data in calculation process are not exactly the same, it is necessary to unify the data interval before the real operation. The index is planned to the interval [0, 1] as far as possible, and the unified processing method of numerical data is given. Qos(q)_i is used to express the value of q corresponding to the i node. Qos(q) represents the aggregate value of the index q. The accurate index of polymerization value is Qos(Acc), and it is converted to Qos^,(Acc). The smaller the value, the better the performance of the data is. In order to check the effectiveness of the results, the value of w_t, w_c, w_a, w_l is 0.25. In the actual calculation process, the weight of the model is corrected according to the weight of the user. Therefore, the expression of the fitness function can be transformed into:

$\begin{matrix} f (x) & = & min (w_{t} \times Qos (Time) + w_{c} \times Qos (Cos t) \\ + w_{a} \times {Qos}^{,} (Acc) + w_{l} \times Qos (Lost)) . \end{matrix}$ (20)

4 Optimization algorithm calculation and model evaluation results

The optimized particle swarm algorithm is tested and analyzed. Basic environmental of test is personal computer, and the operating system is Windows 8. Myeclipse is selected as IDE. Network operating environment and platform is JAVA. Simulation and calculation platform for algorithm is Matlab. The constraints in the process of test are four indicators. The final weight value of the index is 0.25. Figure 7 shows comparative analysis results between the optimized particle swarm optimization and linear programming methods. The total number of particles and the size of the test are about 100, so the iteration number of the algorithm is set to 100. In Fig. 7, abscissa represents the number of alternative logistics service examples for different nodes of basic task logic structure, and the vertical axis represents the time that an algorithm performs at completion time. It can be seen from Fig. 7 that with the gradual increase of the number of alternative service logistics service set, when the number increases to 60, the time needed by linear programming algorithm decreases instantaneously. It can also be understood that the time increment required by the algorithm increases instantaneously. In contrast, the time increment required by the optimized particle swarm algorithm is relatively flat. Therefore, compared with the execution time of the two algorithms, the execution speed comparison analysis shows that the optimized particle swarm optimization algorithm is superior to the linear programming algorithm.

Fig. 7

Experimental results of algorithm operation efficiency.

Figure 8 shows performance comparison test results corresponding to the two algorithms. The total number of test set particle swarm experiment is 100. The abscissa in Fig. 7 represents the number of iteration and the vertical axis represents the search results. It can be seen from Fig. 7 that with the gradual increase of the number of abscissa iterations, both the optimization method and the genetic algorithm can account for the optimal solution, and can accurately confirm the optimal solution is 0.86. But from the number of iterations, when the number of iterations is less than 100 times, the number of iterations of the optimized particle swarm algorithm searches to the optimal solution is the least. And as long as the number of iterations meets 100 times, it can achieve the iterative optimal solution requirements. The genetic algorithm has certain birth defects in the iterative search. Therefore, it can be seen from the results in Fig. 8 that compared with the genetic algorithm, the particle swarm optimization algorithm has better searching ability.

Fig. 8

Algorithm performance comparison test results.

Figure 9 shows the reliability experimental result of the optimized particle swarm optimization algorithm. The size of the particle algorithm in the experiment is 300. The abscissa in Fig. 9 represents the number of iterations, and the vertical axis represents the probability of possible optimal solutions obtained after calculation. The results in Fig. 9 show that when the number of iterations is small, the probability is low when the optimized particle swarm algorithm searches for the optimal solution. When the number of iterations of the particle swarm optimization algorithm is 40 times, the probability of the optimal solution can be close to 96%. And when the iteration number is the same, the probability of finding the optimal solution of genetic algorithm is less than 95%. When the number of iterations reaches 50 times, the probability of searching the optimal solution by genetic algorithm has not reached 100%. At this time, the optimized particle swarm optimization algorithm has reached 100%. From the analysis of the searching ability of the optimal solution, the particle swarm optimization algorithm is superior to the genetic algorithm.

Fig. 9

Experimental results of algorithm reliability.

The optimal particle swarm algorithm has the following advantages and characteristics. When the number of logistics combination service is relatively large, the efficiency of the algorithm is relatively high in the calculation process, and the algorithm shows a strong convergence. When the number of iterations is low, the global optimal solution can be obtained. In contrast, the genetic algorithm needs repeated iterative calculation to find the optimal solution. And the genetic algorithm is easy to fall into the “premature” state, which can’t be explicitly guaranteed to find the optimal solution every time. The optimized particle swarm optimization algorithm in cloud environment proposed in this paper can be well integrated into the logistics service composition algorithm. And the optimized particle swarm algorithm can avoid the algorithm search into the deadlock of local optimization. This can be more convenient to deal with the basic model of logistics portfolio services.

5 Conclusion

In this paper, based on the combination of Web service technology and international logistics characteristics in cloud environment, the characteristics of cloud service technology of international logistics were fully discussed. The service and evaluation of logistics were analyzed. Based on the Qos logistics service calculation model, the optimal service calculation model of the combined logistics service model was designed. The quality of logistics service combination was evaluated. According to the specific needs of the model, the weights of expert evaluation index and user evaluation index were added to balance the evaluation objectivity of the combined logistics service model. In order to solve the disadvantages of the traditional algorithm better, the optimized particle swarm algorithm was optimized, and the fitness function was added to verify the scientificity and effectiveness of the optimized particle swarm optimization algorithm. In comparison experiments, the results showed that the efficiency of optimized particle swarm algorithm to find the optimal solution was high and the speed was fast. And the optimized particle swarm algorithm overcame the “premature” feature of the traditional algorithm, and successfully avoided the trapped into the local optimal solution calculation state. At the same time, when the number of iterations of the global optimal solution was low, the completion probability was close to 100%. The improved particle swarm optimization algorithm had obvious advantages. Therefore, more indicator demands should be proposed to future users in the logistics network services. Therefore, it was necessary to use logical process modeling to solve the combinatorial circulation scheme of logistics.

Footnotes

Acknowledgments

The research was founded by China (Xi’an) Institute for Silk Road Research Program No. 2016SZ06 entitled: “Central Asia –Horgos Free Trade Zone –Xi ‘an International Harbor Area: One Belt and One Road international logistics chain” and International Science and Technology Cooperation and Exchange Program (Department of science and technology of Shaanxi province) No. 2016KW-065 entitled: “The Construction of logistics collaborative innovation service platform along the Silk Road nations (China & Kazakhstan)”.

References

Tao ,

G.J.

Sheng ,

Center , et al., Web service composition based on modified particle swarm optimization, Chinese Journal of Computers36(5) (2013), 1031–1046.

Ma ,

Xu , et al., Two-echelon logistics distribution region partitioning problem based on a hybrid particle swarm optimization-genetic algorithm, Expert Systems with Applications42(12) (2015), 5019–5031.

A. A. A.

Esmin and

R.A.

Coelho , Consensus clustering based on particle swarm optimization algorithm, IEEE International Conference on Systems, Man, and Cybernetics, IEEE Computer Society (2013), 2280–2285.

Liao ,

Liu ,

Zhu , et al., Accurate sub-swarms particle swarm optimization algorithm for service composition, Journal of Systems & Software90(1) (2014), 191–203.

Wu ,

Ke ,

Lei , et al., Web service composition recommendation based on modified particle swarm optimization, Journal of Huazhong University of Science & Technology42(10) (2014), 69–73.

Hua ,

Hu and

Yuan , Research optimization on logistics distribution center location based on adaptive particle swarm algorithm, Optik - International Journal for Light and Electron Optics127(20) (2016), 8443–8450.

Zou ,

Tao ,

Jiang , et al., A new approach for data processing in supply chain network based on FPGA, International Journal of Advanced Manufacturing Technology84(1-4) (2016), 249–260.

Liao ,

Liu ,

Song , et al., Markov chain-like model for prediction service based on improved hierarchical particle swarm optimization cluster algorithm, International Journal of Software Engineering&Knowledge Engineering26(04) (2016), 653–674.

T.J.

Chen ,

L.L.

Wu ,

J.J.

Liang , et al., Research and analysis on ionospheric composition based on particle swarm optimization, International Conference on Intelligent Computing Theories and Technology, Springer-Verlag (2013), 596–604.

10.

Lu ,

Hu ,

Wang , et al., A hybrid algorithm based on particle swarm optimization and ant colony optimization algorithm, Smart Computing and Communication, Springer International Publishing (2016), 22–31.

11.

J.X.

Wang and

Li , WSN coverage enhancement algorithm based on particle swarm optimisation, Advanced Materials Research945-949(2) (2014), 2386–2393.

12.

Xiao and

Li , Embedded database query optimization algorithm based on particle swarm optimization, Seventh International Conference on Measuring Technology and Mechatronics Automation, IEEE (2015), 429–432.

13.

Zhou ,

C.Z.

Chen ,

X.M.

Sun , et al., Research on blind source separation algorithm based on particle swarm optimization, Advanced Materials Research989-994 (2014), 1566–1569.

14.

Cao ,

Li and

Liu , An on-demand service composition method based on trustworthy quality of service, Journal of Xian Jiaotong University47(2) (2013), 131–138.

15.

Wang ,

Zhang ,

Li , et al., Particle swarm optimization algorithm based on two swarm evolution, Control and Decision Conference, IEEE (2015), 120–1204.

16.

O.J.

Alejo ,

J.M.

Fernandez-Luna and

J.F.

Huete , RankPSO: A new L2R algorithm based on particle swarm optimization, Journal of Multiple-Valued Logic & Soft Computing23(1) (2014), 1–34.

17.

Gong ,

Zhang ,

X.U.

Ke , et al., A transportation routing problem of emergency materials based on particle swarm optimization algorithm, Journal of Chongqing Normal University11(1) (2015), 1–18.

18.

Yang , Learning resource recommendation method based on particle swarm optimization algorithm, Journal of Computer Applications27(27) (2014), 2257–2263.

19.

Li ,

Ding and

Li , A novel naive bayes classification algorithm based on particle swarm optimization, Open Automation & Control Systems Journal6(1) (2014), 747–753.

20.

Cheng , Multi-target tracking algorithm based on improved simplified particle swarm optimization, Journal of Chongqing Normal University31(02) (2016), 486–488.