### Research on the Internet of Things Based on Ant Colony Optimization Algorithm

#### Yibin Hou 1* , Jin Wang 1

1 School of Software Engineering, Department of Information, Beijing University of Technology, Beijing, China

### Abstract

The purpose of this paper is to prove that the ant colony algorithm is an excellent mathematical modeling method and improve the production efficiency of the single drilling machine. The methods of mathematical induction and mathematical deduction and mathematical hypothesis are commonly used mathematical methods in scientific research. The ant colony algorithm to solve the TSP problem: algorithm design ideas: using the standard ant colony algorithm or its improved to achieve a traveling salesman problem TSP, find the shortest distance of the 51 City, the number of iterations is 1000 times, the final output of the optimal solution. Algorithm flow: (1) initialize ant colony: initialize ant colony parameter, set ant number, ant put in 51 vertices, initialize path pheromone. (2) Ant action: the ants leave their paths by the ants in front of the pheromone and their own judgments to complete a loop path. (3) Releasing pheromones: the path to releasing ants through a certain percentage of pheromones. (4) The evaluation of ants: the fitness is evaluated according to the objective function of each ant. (5) If the shortest path condition is satisfied, the optimal output is obtained. Otherwise, the algorithm continues. (6) Pheromones evaporate: pheromones continue to dissipate over time. The result of this paper is that the ant colony algorithm has high accuracy and efficiency; the TSP problem can be solved, to improve the production efficiency of the single drilling machine. The conclusion of this paper is that ant colony algorithm is an excellent algorithm, the TSP problem can be solved, for example, can improve the production efficiency of the single drilling machine.

### Keywords

Internet of Things, Colony Optimization Algorithm, TSP Problem, the Production Efficiency of the Single Drilling Machine

### 1. Introduction

The Internet of things is a kind of Internet, and all the problems in this article are studied under the Internet of things. The methods of mathematical induction and mathematical deduction and mathematical hypothesis are commonly used mathematical methods in scientific research. Artificial intelligence includes machine learning, and machine learning includes ant colony algorithm [1-10].

1.1. The Ant Colony Algorithm to Solve TSP Problem

The ant colony algorithm to solve the TSP problem:algorithm design ideas:using the standard ant colony algorithm or its improved to achieve a traveling salesman problem TSP, find the shortest distance of the 51 City, the number of iterations is 1000 times, the final output of the optimal solution. Algorithm flow:(1) initialize ant colony:initialize ant colony parameter, set ant number, ant put in 51 vertices, initialize path pheromone. (2) Ant action:the ants leave their paths by the ants in front of the pheromone and their own judgments to complete a loop path. (3) Releasing pheromones:the path to releasing ants through a certain percentage of pheromones. (4) The evaluation of ants:the fitness is evaluated according to the objective function of each ant. (5) If the shortest path condition is satisfied, the optimal output is obtained. Otherwise, the algorithm continues. (6) Pheromones evaporate:pheromones continue to dissipate over time. Basic design idea:(L) pre initialization of pheromone intensity and ant tabu list. Ants follow the rules in the tabu list of certain probabilities and arrive at the nodes at the next choice until a legitimate path is formed. (2) Calculatethe length of the path generated by each ant, and the path length is the sum of the length of each path. (3) Update the pheromone on each side. Each side of the first pheromone volatile operation, and then, according to the path length generated by ants, to get the release of acne by pheromones. (4) when all the ants have completed the pheromone update operation, record the current shortest paths, and the tabu list and the pheromone value added in the △( ttl) is initialized, and go to step 2. And so on, until the end of the algorithm is satisficannoth as the solution can not be further improved or reached the predetermined number of cycles. First, set up an ant class, and then set up a TSP class, call the ants in TSP class variables and methods, and then define the main function, in the main function calls the TSP class variables and methods [1-10].

1.2. The Artificial Ant Colony Algorithm for TSP Problem

In the artificial ant colony algorithm of TSP problem, it is assumed that the M, ant moves between the adjacent nodes of the graph, thus the solution of the problem can be obtained asynchronously and asynchronously. The probability of a step shift for each ant is determined by the two parameter on each edge of the graph:(1) Pheromone values are also called pheromone traces. (2) Visibility, that is, a priori value. There are 2 kinds of information update in a way, is volatile, which is all the information on the path for a certain ratio of reduction, process simulation of natural pheromone volatile over time;the two are to enhance the value, to "good"(ants through side) increase pheromone. The downward movement of a target is realized by a random principle, is the use of the storage node information, the next step is to calculate the probability of a node, and according to the probability of achieving a step by the reciprocating movement, more and more close to the optimal solution. Ants in the search process, or find a solution, will evaluate the optimization of the solution or part of the solution, and the evaluation information stored in the relevant connection pheromone. Ant colony algorithms are used to find the shortest distance between 51 cities [11-20]. The flowchart of the ant colony algorithm is shown in Figure 1 .

Figure 1. Flow chart of ant colony algorithm.

1.3. The Idea of Algorithm Design

Using the standard ant colony algorithm or its improved traveling salesman problem TSP, find the shortest distance between 51 cities, the number of iterations 1000 times, and finally output the optimal solution. The basic flow of the algorithm is:in the artificial ant colony algorithm for the TSP problem, it is assumed that the M, ant moves between the adjacent nodes of the graph, thus the solution of the problem can be obtained asynchronously and asynchronously. One step of each ant's transition probability is determined by two parameters on each edge of the graph: (1), pheromone values are also called pheromone traces. (2) visibility, that is, a priori value. There are 2 kinds of information update in a way, is volatile, which is all the information on the path for a certain ratio of reduction, process simulation of natural pheromone volatile over time;the two are to enhance the value, to "good"(ants through side) increase pheromone. The downward movement of a target is realized by a random principle, is the use of the storage node information, the next step is to calculate the probability of a node, and according to the probability of achieving a step by the reciprocating movement, more and more close to the optimal solution. Ants in the search process, or find a solution, will evaluate the optimization of the solution or part of the solution, and the evaluation information stored in the relevant connection pheromone [21-30].

1.4. Algorithm Flow

Algorithm flow:(1) Initializing ants:initializing ant colony parameters, setting ants quantity, placing ants on 51 vertices, initializing path pheromone. (2) Ant movement:the ant chooses the path according to the pheromone left by the ant in front and its own judgment, and completes a cycle. (3) Release pheromone:release the pheromone in a certain proportion to the path of the ants. (4) Evaluate the ant colony:evaluate the fitness of each ant, according to the objective function. (5) If the condition of the shortest path is satisfied, the output optimal solution is obtained. Otherwise, the algorithm continues. (6) The pheromone pheromone volatilization:as time continues to dissipate.

1.5. Ant Colony Optimization Algorithm

Ant colony optimization algorithm: (1) A group of ants starts randomly from the starting point, meets food, holds food, and returns along the road. (2) Ants on the way back and forth, leaving pheromone signs on the road. (3) The pheromone will gradually evaporate over time (usually available as a negative exponential function or released at a certain rate). (4) From the nest of ants, the path selection and the probability of each path pheromone concentration is proportional to the. Note:the same principle can be used to describe the foraging situation of multiple food sources by ant colonies. The framework of general ant colony algorithm has three components:A: the activity of ant colony;B:the volatilization of pheromone;C: the enhancement of pheromone;the transition probability formula and the pheromone update formula are mainly reflected in the previous algorithm [31-40].

1.6. Basic Design Idea

Basic design idea:

(t):t times the number of ants in the city of i;

m:The total number of ants in an ant colony, m=;

:Pheromone intensity on edges (i, j);

:Visibility on sides (i, j);

:The distance between city i and city j;

:The probability of the transfer of ant k from urban i to urban j.

Pre initialization pheromone intensity on each side, and tabu list of ants. In accordance with certain rules of probability, the ants select the next node to be reached under the restriction of the tabu list, and eventually form a legitimate path. (2) Pre initialization pheromone intensity on each side, and tabu list of ants. In accordance with certain rules of probability, the ants select the next node to be reached under the restriction of the tabu list, and eventually form a legitimate path. (3) Update the pheromone on each side. Each side first carries out pheromone volatilization operation, and then obtains the pheromone released by ants according to the path length generated by each ant. (4) When all ants have completed the pheromone update operation, the current shortest path is recorded, and the tabu table and pheromone values are added ( ttl) proceed to initialization and go to step (2). And so on, until the end of the algorithm is satisfied, such as the solution cannot be further improved or reached the predetermined number of cycles. First, set up an ant class, and then set up a TSP class, call the ants in TSP class variables and methods, and then define the main function, in the main function calls the TSP class variables and methods.

### 2. Methods

A hole is a printed circuit board (also known as printed circuit board) one of the important components of the processing cost holes usually accounted for 30% to 40% of the cost of the system board, punching machine is mainly used in the drilling process in the manufacture of printed circuit board, the problem is to improve the production efficiency of a certain kind of punch.

The production efficiency of drilling depends mainly on the following two aspects: (1) the drilling time of the through hole, which is determined by the manufacturing process, assumes that the operation time for each hole is the same.(2) punch in the processing operation, bit travel time (or walk), assuming the drill only horizontal and vertical motion. A company shall produce a batch of printed circuit boards of large quantity and of the same specifications, and the data for the through-hole coordinates are as follows:please design an optimum through-hole plan for the company.

Data figure as shown below in Figure 2 , (x, y) represents the center coordinate data of the hole;the Center coordinates data of printed circuit board as shown in Table 1 .

Figure 2. Data figure.

Table 1. Center coordinates data on printed circuit board.

 x y x y x y x y x y 288 149 32 121 48 73 64 21 260 37 288 129 32 113 56 73 72 25 260 45 270 133 40 113 56 81 80 25 260 53 256 141 56 113 48 83 80 25 260 61 256 157 56 105 56 89 80 41 260 69 246 157 48 99 56 97 88 49 260 77 236 169 40 99 104 97 104 57 276 77 228 169 32 97 104 105 124 69 276 69 228 161 32 89 104 113 124 77 276 61 220 169 24 89 104 121 132 81 276 53 212 169 16 97 104 129 140 65 284 53 204 169 16 109 104 137 132 61 284 61 204 169 16 109 104 137 132 61 284 61 196 169 8 109 104 145 124 61 284 69 188 169 8 97 116 145 124 53 284 77 196 161 8 89 124 145 124 45 284 85 188 145 8 81 132 145 124 37 284 93 172 145 8 73 132 137 124 29 284 101 164 145 8 65 140 137 132 21 288 109 156 145 8 57 148 137 124 21 280 109 148 145 16 57 156 137 120 9 276 101 140 145 8 49 164 137 128 9 276 93 148 169 8 41 172 125 136 9 276 85 164 169 24 45 172 117 148 9 268 97 172 169 32 41 172 109 162 9 260 109 156 169 32 49 172 101 156 25 252 101 140 169 32 57 172 93 172 21 260 93 140 145 8 49 164 137 128 9 276 93 148 169 8 41 172 125 136 9 276 85 164 169 24 45 172 117 148 9 268 97 172 169 32 41 172 109 162 9 260 109 156 169 32 49 172 101 156 25 252 101 140 169 32 57 172 93 172 21 260 93 132 169 32 65 172 85 180 21 260 85 124 169 32 73 180 85 180 29 236 85 116 161 32 81 180 77 172 29 228 85 104 153 40 83 180 69 172 37 228 93 104 161 40 73 180 61 172 45 236 93 104 169 40 63 180 53 180 45 236 101 90 165 40 51 172 53 180 37 228 101 80 157 44 43 172 61 188 41 228 109 64 157 44 35 172 69 196 49 228 117 64 165 44 27 172 77 204 57 228 125 56 169 32 25 164 81 212 65 220 125 56 161 24 25 148 85 220 73 212 117 56 153 16 25 124 85 228 69 204 109 56 145 16 17 124 93 228 77 196 101 56 137 24 17 124 109 236 77 188 93 56 129 32 17 124 125 236 69 180 93 56 121 44 11 124 117 236 61 180 101 40 121 56 9 124 101 228 61 180 109 40 129 56 17 104 89 228 53 180 117 40 137 56 25 104 81 236 53 180 125 40 145 56 33 104 73 236 45 196 145 40 153 56 41 104 65 228 45 204 145 40 161 64 41 104 49 228 37 212 145 40 169 72 41 104 41 236 37 220 145 32 169 72 49 104 33 236 29 228 145 32 161 56 49 104 25 228 29 236 145 32 153 48 51 104 17 228 21 246 141 32 145 56 57 92 9 236 21 252 125 32 137 56 65 80 9 252 21 260 129 32 129 48 63 72 9 260 29 280 133

2.1. Problem Analysis

Problem analysis: This will improve the production efficiency of the machine and make an analysis, the production efficiency refers to the unit of time production capacity, processing efficiency. So in order to improve the production efficiency of drilling, we can make the travel time and the total drill tool change time as short as possible, the production efficiency will be higher.

For a single drill, we first draw the distribution of all points to determine that the diagonal of the circuit is approximately, and that the speed of travel is rough and that the time of travel is not long. Considering the moving speed and the conversion time of the tool, it is found that the conversion time of the tool is much longer than the travel time, so we want to achieve the shortest time when the tool's conversion time is the best. Therefore, we use an ant colony algorithm to compute the shortest distance between the shortest distance and the least number of tool conversions, and compare the two, and get the optimal results.

2.2. Model Hypothesis

Model hypothesis:

(1) The drilling time of a single through the hole is determined by the manufacturing process. In order to simplify the problem, it is assumed that the drilling time for the same pass is the same.

(2) In order to calculate the travel expenses, need to calculate the travel time, in order to simplify the problem, assuming the punch travel is uniform motion.

(3) It is assumed that the change over time of the tool is the same for different pass processing operations;

(4) Between the two hole distance calculation, to simplify the problem, the drill bit is a particle hypothesis punch.

(5) In order to avoid the touch and interference between the drills, it is assumed that the distance between the two drills is not less than 3cm.

2.3. Model Establishment

Model establishment:A single bit model:for a given size of the hole, adjust the corresponding tool, from the point of the tool knife along the shortest distance path, moving from one hole to another hole, the hole until all objects is processed, processing the other hole then the next size conversion tool, such an arrangement. Describe the problem as the following optimization model:

(1) Variable design. A collection with n holes ……set upRepresents any two holes in a collection,Represents the distance between two holes in a collection,M is the total distance traveled.

(2) Objective function. You need to find a non repeating whole arrangement in the hole collection

order , find the minimum value of M.

(3) Constraints:the machining path starts from one hole, processes only one hole at each hole, traverses each hole, and finally goes back to the starting point, including tool switching.

(4) Optimization algorithm:ant colony algorithm.

### 3. Results

Results analysis of single drill bit:For the single bit mode of production, way of using a drilling tool for drilling holes corresponding to all a tool of the corresponding hole drilling in conversion tool, the line of work by using ant colony algorithm, data processing by MATLAB. (In order of order Cutting tool:defghabcf, Pass:DGDIJFGHFACBCEIJEG.

The following table shows the stroke of each tool (unit: 104mil), the stroke of each tool as shown in Table 2 .

Table 2. The stroke of each tool.

 tool d e f g h a b c f Trip 5.9257 5.9527 4.6661 3.2375 3.0941 12.586 11.573 11.123 4.8309 time/s 8.361821 8.399921 6.584386 4.568472 4.366119 17.76024 16.33079 15.69579 6.816937

According to the above table, under the optimal line condition of single drill operation, the total length of various tool operations can be as follows:62.989e+004 mil. The following is the point of view we have set for each tool (1/100mil), the point of view we have set for each tool as shown in Table 3 .

Table 3. The point of view we have set for each tool.

 d e f g h a b c f x -267400 -257400 -222047 -301300 -311300 -202800 -197600 -202800 199800 y 184518 184518 100000 84300 74300 190200 180800 190200 203200

### 4. Discussion

The distance from the previous table and the distance traveled for the selection of 1.2471e+004mil, plus the distance of advance, shows that the total distance of the route is Shortest_Length= 64.2361e+ 004 (mil) = 16315.97 (mm). The speed of movement of all bits is the same, and the drilling time of all bits of the drill is

12*18=216 (s). Timing:job travel time = shortest distance / bit travel speed =90.64427 (s),

Total operating time = drill travel time + drill conversion time =306.64 (s),

Cost:travel cost = travel time * travel cost =978.9594 (yuan),

Job conversion cost = conversion times * conversion cost =25.2 (yuan),

Total operating cost = travel cost + job transfer cost =1004.159 (yuan),

That is to say, this method will cost 306.64 (s) and cost 1004.159 (yuan) when needed[43-47].

### Acknowledgments

Thanks to the National Natural Science Foundation of China (No:61203377, 60963011, 61162009), the work was supported by scholarships and fellowships for tutors, doctoral scholarships and fellowships from the Software Institute of Beijing University of Technology and the Faculty of information. Here, thank you very much for Hou Yibin tutor support and help and guide the care, thanks to Mentor Professor Hou Yibin zhiyuzhien. Thank you very much for your mentors, providing good research platform and environment, providing opportunities for self - exercise, building and providing self - value platform, providing strong help and support, and teaching research methods. Thank you very much for the teacher's knowledge and careful guidance, provides a good experimental environment for students, Hou teacher plays a key role in my study and research, and the growth of each character every step cannot do without the guidance of the teacher hou. Hou's academic attainments are very high, both theoretical and practical. Thanks to Professor Hou Yibin for my work, learning, research, life, all aspects of care, guidance and help, in his proper guidance, it is easier to write a paper, published many (including ISTP, EI, SCI conferences and journals) and IEEE papers. Hou's domestic and international perspective and keen insight and academic in Beijing and the Beijing University of Technology and social work methods and in-depth understanding of teamwork and selfless work attitude and painstaking work style and painstaking patience to cultivate students advocating morality, benefited me a lifetime. Hou has worked hard for my study and research. This is an inspiring mentor; he is generous in the atmosphere.

### References

[1] Golden sea soluble; Zhou Su; Mencius. Experiment method reform assembly language programming teaching [J]. Computer education, 2007, 01S, 34-37.

[2] Li Jingxia; Hou Zifeng. Application of Web service composition modeling and application of [J]. Computer based on colored Petri nets, 2006, 1, 22-27.

[3] Li Pei. Fully digital emulation automatic generation control system [D]. Lanzhou University of Technology, 2006.

[4] Li Yun; Wang Qiangjun. Read Mr. Zhao Yuanrens article the meaning of language and its acquisition [J]. Language application, 2001, 4, 70-75.

[5] Bao Xinzhong; Zhang Jianbin; Liu Cheng. [J]. China scientific management, to determine the weight of conditional information entropy based on rough set. 2009, 17(3), 131-135.

[6] Gao Zhijun; Yan Guozheng; Ding Guoqing; et al. Research status and development of multi robot coordination and cooperation system and development of [J]. Optical precision engineering, 2001, 9 (2).

[7] Sun Xida. Research on Data Mining Based on BOM in discrete manufacturing industry [D]. Kunming University of Science and Technology, 2004.

[8] Zhou Hongmei. An overview of the main algorithms for intelligent computing [J]. Computer knowledge and technology, 2010, 9, 71

[9] Ma Chi. Face recognition algorithm research and implementation [D]. Beijing Jiaotong University, 2007.

[10] Cao Zhaolong; Wan Yong. SVM algorithm and the multi class letter image recognition using [C]// twenty-fifth Chinese Control Conference (2) 2006.

[11] Mao ting. Research on the recognition of Chinese proper nouns [D]. Dalian University of Technology, 2006.

[12] Lin Peng. Research and implementation of face detection based on Adaboost algorithm [D]. Xian: School of computer science, Xian University of technology, 2007.

[13] What is the relationship between machine learning, statistical analysis, data mining, neural networks, artificial intelligence, pattern recognition? - How time flies! - blog Channel - CSDN.NET,

Available online: http://blog.csdn.net/u011023036/article/details/49 (accessed on 1 December 2017).

[14] Overview of the relationship between artificial intelligence, machine learning, data mining, neural network and other disciplines (collection compilation) -- love programming, Available online: http://www.w2bc.com/article/181993 (accessed on 1 December 2017).

[15] Hope learning Available online: http://scholar.google.com/schhp? Hl=zh-CN&as_sdt=0c... (accessed on 1 December 2017).

[16] Machine learning, statistical analysis, data mining, neural network, artificial intelligence, pattern recognition -- look up at the starry McDull blog CSDN.NET. Available online: http://blog.csdn.net/u012116229/article/details/24 (accessed on 1 December 2017).

[17] Overview of the relationship between artificial intelligence, machine learning, data mining, neural network and other disciplines (collection compilation) -- love programming, Available online: http://www.w2bc.com/article/181993 (accessed on 1 December 2017).

[18] Machine learning, statistical analysis, data mining, neural network, artificial intelligence, pattern recognition. Available online: http://www.itpub.net/thread-1861681-1-1.html (accessed on 1 December 2017).

[19] Zhang Yangsen; Huang Gaijuan; editor. A tutorial on artificial intelligence, [M]. higher education press, 2017

[20] Sleep with big data, TIPS! – Pop society . Available online: http://www.liuxingshe.com/qt/1038575.html (accessed on 1 December 2017).

[21] Pan Zhijie; Ren Saijun; Chen Yourong. A single digital character recognition algorithm of BP neural network to realize [J]. Computer knowledge and technology, based on the 2008, 29, 86.

[22] Hu Zhenxin; Li Hong; Guo Zehua. Research on multi class classification methods of support vector machines [J]. Modern electronic technology, 2011, 34(13), 165-167.

[23] Lin Bihua; Gu Xingsheng. Research on differential evolution algorithm and least squares support vector machine modeling based on [J]. Journal of chemical industry and engineering, 2008, 59(7).

[24] Li Siguang; Zhou Xuemei. Machine learning based intrusion detection system [J]. Silicon Valley, 2008, 8, 39-40.

[25] Ma Jie. Research on stock market forecasting based on support vector machine (SVM) [D]. Tianjin University, 2006.

[26] Liu Jingli. Research and application of robust least squares support vector machine [D]. Hefei: School of management of University of Science & Technology China, 2010.

[27] Yang Banghua. Spontaneous brain computer interface technology and identification of EEG signal [D], 2006.

[28] Liu Jian, schaemon. Support vector machine classification algorithm based on K nearest neighbor decision and Simulation of [J]. Ship Electronic Engineering, 2009, 3, 88-91.

[29] Sun Zhihui. Research and application of intelligent recognition method based on Hilbert spectral feature extraction [D]. Dalian University of Technology, 2009.

[30] Chang Guo Feng. Prospect, the networking and the Internet technology between [J]. 2014, 11, 31.

[31] Chao H.Y.; Cao Y.C.; Chen Y.Q. Autopilots, Y, H 2010, 8 (1):

[32] Cui Xiumin; Wang Weijun; Fang Zhenping. Small development status analysis and related problems of M. UAV flight mechanics, 2005, 23(1), 14-18.

[33] Eagle. Chinese Navy UAV application prospect of Shipborne weapons [J]. 2007, 9, 43-49.

[34] Li Jiyu; Zhang Tiemin; Peng Xiaodong. Application of UAV [J]. Research of Agricultural Mechanization in farmland information monitoring system in the 189-192. 2010 32

[35] Aijomandi M.; Agostino S.; Mammone M.; et al. Classification of Unmanned.

[36] Qin Wu; Zhang Aihua; Li Jin. Small UAV [J]. Flying missile in the United States, 2007, 2, 22-25.

[37] Fan Chengchen; Han Jun; Xiong Zhijun; so on. The current situation and application of unmanned aerial vehicle remote sensing technology and application of [J]. Surveying and Mapping Science, 2009, 34(5), 214-215.

[38] Wang Jianping; Yan Zejing. On the application of UAV in the inspection of transmission lines [J]. China Electric Power Education, 2013, 6, 229-230.

[39] Lining. Air traffic management of UAV system [J]. China civil aviation, 2010, 9, 48-51.

[40] Xu Zhiqiang; Jiang Xudong; so on. The application of UAV in the field of earthquake field exploration [J]. International earthquake dynamics, 2012, 6, 204.

[41] Van Chunwei. Things of the legal protection of intellectual property rights of [D]. Shanxi University, 2013.

[42] Zhang Yong. Research and application of intelligent equipment intelligent management system [D]. North China Electric Power University, 2012.

[43] Jiang Qiyuan and so on, mathematical model (Fourth Edition), Beijing: Higher Education Press, 2011.

[44] Han Ming; Wang Jiabao; so on, mathematics experiment, Shanghai: Tongji University press, 2009.

[45] Luo W.C. mathematical modeling case selection, Chengdu: Southwest Jiao Tong University press, 2007.

[46] Li Shiyong; Ken Chan; ant colony algorithm and its application, Harbin: Harbin Institute of Technology press, 2004.

[47] Zeng Feng. Printed circuit board (PCB) design and production, Beijing: Electronic Industry Press, 2002.

[48] Cold Huang. A number of studies on ant colony optimization algorithm [D]. Jilin University, 2016.

[49] Xia Xiaoyun; Zhou Yu. Research progress on the theory of ant colony optimization (ACA), [J]. Journal of intelligent system, 2016.

[50] Wang Zhaoyuan. Research on Optimization of network coding resource based on Ant Colony Optimization (ACO) [D]. Southwest Jiao Tong University, 2016.

[51] Ge Hongwei; Yuan Yunhao; Zhou Niao; Su Shuzhi. Optimization of [J]. Computer engineering and science continuous ant colony based on artificial bee colony, 2016 (6).

[52] Zhao Shian. Analysis and Research on convergence of ant colony optimization algorithm [J]. Modern electronic technology, 2017, 10.