APPLICATION SOFTWARE DESIGN FOR OPTIMIZATION CABLE INSTALLATION WITH ANT COONY METHOD

Transcription

1 APPLICATION SOFTWARE DESIGN FOR OPTIMIZATION CABLE INSTALLATION WITH ANT COONY METHOD Joni Cukri Binus University, Jalan KH. Syahdan No. 9 Palmerah, Jakarta 11480, Indonesia joni.cukri@yahoo.co.id ABSTRACT Cabling installation is one of the work for housing developers to build house. By now, they haven t a proper plan for cable installation, foreman or worker leader nowdays build randomly and nonconsecutive that make the installation worse, because workers had to cut and connect the cable is installed with new cables. It also creates unefficiently use of cables which caused harm to the company. Therefore, the authors was made to make a plan for them to make an efficient cable installation planning by using Ant Colony Method. This method can give a solution which the cable installation sequences are done in every houses. The range of the cable is also included in this method, so that the developer can buy the exact length of the cable that they needed. The results of the method was tested well in making sequences from one house to another by shortest part. The system also can display the total distance pipeline developer that can help provide the necessary requirements. Keywords: optimization, ant colony method

2 ABSTRAK Pemasangan kabel menjadi bagian pekerjaan bagi developer perumahaan dalam membangun perumahan. Saat ini proses pemasangan kabel tidak ada perencanaan yang matang, mandor atau pimpinan pekerja memasang kabel secara acak dan tidak berurutan sehingga kabel yang terpasang tidak bagus, karena pekerja harus memotong dan menyambung kabel yang sudah terpasang dengan kabel yang baru terpasang. Hal ini juga menimbulkan tidak efisien penggunaan kabel yang menimbulkan kerugian bagi perusahaan. Tujuan Skripsi ini adalah penulis ingin merancang sistem pencarian hasil optimal pemasangan kabel menggunakan metode ant colony. Metode ant colony mampu memberikan solusi dengan memberikan urutan pemasangan kabel ke setiap rumah dengan baik. Melalui sistem ini juga dapat diketahui jarak pemasangan kabel yang akan dipasang, sehingga memudahkan developer untuk menyediakan kabel sesuai kebutuhan. Hasil dari sistem optimasi pemasangan kabel menggunakan metode ant colony telah teruji berjalan dengan baik dalam menentukan urutan pemasangan kabel berdasar jarak terpendek. Pengujian dengan menggunakan input dari jarak antara rumah ke rumah dan dilakukan kalkulasi dengan menggunakan ant colony tersebut. Kata kunci: optimasi, Metode Ant Colony INTRODUCTION For residential contractors, wiring be part of the job. In conducting the necessary cabling installed the sequence planning in advance, in order to work efficiently in the process of time, labor and material costs. In addition, the results of the neat and sequential. Currently there are no cabling planning, installing cable foreman only on homes that are ready inhabited. It certainly is not efficient because of wiring be sequential and not neat. In addition, if the distances are different from each other has always been a barrier at the time of installation, so it is not cost efficient for the company. Based on the authors provide a solution to these problems by making the wiring optimization applications using ant colony. Ant Colony method inspired by foraging behavior of ants, which use pheromones to mark the path their way to the food source ( Dorigo, Gambardella, 2004). This method works on the trail left by other ants or commonly called pheromones. With the use of ant colony can be found between the house wiring sequence. Expected with the use of this method can provide solutions and benefits for the company Ant Colony Optimization ( ACO ) was adopted from the behavior of ant colonies are known as ant system ( Dorigo, M. and Gambardella, L., 1996). Naturally ant colony is able to find the shortest route on the way from the nest to a food source places. Colonies of ants can find the shortest route between the nest and the food source by footprints on the path that has been traversed. The more ants through a track, the more obvious traces of his feet. This will lead to the path traversed ants in small amounts, the longer it will decrease the density of ants that passed, or even will not be passed at all. Instead, the path traversed ants in large quantities, the longer it will further increase the density of ants that passed, or even all of ants going through the trajectory. At each ant ACO placed at all points of the graph ( in this case the point - the point being visited ) which will then move to visit the whole point. Each ant will make each track - each to return to the place where they were initially placed first. If you've

3 reached this state, the ant has completed a cycle ( tour ). Final solution is the shortest path of all paths generated by the ants search. ACO algorithms have been widely applied in various fields which include several issues, namely : 1. Traveling Salesman Problem (TSP), which is looking for the shortest route in a graph using the Hamilton. 2. Quadratic Assignment Problem (QAP), which assigns a number n resources to be placed on a number of m locations by minimizing the cost assignment (assignment ). 3. Job-shop Scheduling Problem (JSP) is also an example of application of Ant Colony Optimization, which is to find a path of n jobs using m number of machines so that all work is completed in the minimum possible time. 4. Vehicle Routing Problem (VRP) 5. Vehicle routing 6. Graph coloring 7. Implementation of the communication network 8. Network routing, dll. Given the principle of an algorithm based on ant colony behavior in finding the shortest traveling distance, ACO is highly appropriate to be applied in solving optimization problems, one of which is to determine the shortest path between the house wiring to the house. In the development of this system, will be logged, data input, and get results with Ant Colony method. METHODS In engineering applications, the analysis is done by looking for the basic theory of the method of ant colony, system design and user interface design and assisted with the supervisor to determine the application design constraints. In the design analysis will be aided by analyzing the problems occurred, analyze and use the solution given data flow, flowchart and use case diagrams. Then for input data that will be needed for testing will be taken randomly. RESULTS AND DISCUSSION Ant Colony Optimization ( ACO ) was adopted from the behavior of ant colonies are known as ant system ( Dorigo, et.al, 1996). Ant complex is able to sense their environment to search for food and then back to the nest by leaving pheromone substance on these routes they take. Pheromone is a chemical derived from endocrine glands and is used by living things to recognize same-sex, other individuals, groups, and to assist the process of reproduction. In contrast to the hormone, pheromone spread outside the body and can only affect and be recognized by other similar individuals ( one species ). This process is known as Pheromone relics stigmery, a process of modifying the environment that not only aims to remember the way back to the nest, but also allows the ant colony communicate with. Over time, however, the trail pheromone will evaporate and will reduce the strength of its appeal. Each ant faster commute through this route, then the pheromone that evaporates less. Vice versa if the ant longer commute through this route, then the pheromone that evaporates more. Philosophy of Ant Colony Algorithm is: 1. Problem solving is the modeling of reality ant colony in search of food 2. Number of alternative solutions to a problem-solving are a number of ants in the colony 3. Population of the number of alternative solutions is each time the colony went in search of food and return to their nests

4 Benefits ant colony algorithm is also used to search for a path / optimal path ( either distance, time and cost ), at a goods distribution network, the algorithm can also be used to find the optimal value of another that occurred in the field, such as a combination of issues, values production and can also be used for forecasting. As he explained how the ants finding the shortest route in the ACO are as follows : ants naturally capable of finding the shortest route on the way from the nest to a food source places. Colonies of ants can find the shortest route between the nest and the food source by footprints on the path that has been traversed. The more ants through a track, the more obvious traces of his feet. This will lead to the path traversed ants in small amounts, the longer it will decrease the density of ants that passed, or even will not be passed at all. Instead the path traversed ants in large quantities, the longer will be growing through the density of ants, or even all of the ants will go through the path ( Dorigo, M., Maniezzo, V., and Colorni, A., 1991a Picture 1. Ants journey from nest to food sources. Picture 1.a above shows two groups of ants that will make the trip. One group called L ie group departed from the left which is an ant nest and another group called the R group departed from the right which is a food source. Both groups of ants from the initial point of departure is in the position of decision-making on which, to be loaded. Ant group L divides two groups again. Partly by road and partly by way of the bottom. This is also true of the ants R. Figure 1.b and 1.c images show that the ant group runs at the same speed by leaving pheromone ( ants footprints ) on the road that has been traversed. Pheromone left by ants - ants through the street has undergone a lot of evaporation because of ants through the road amounted to less than the street below. This is because the distance is longer than the bottom of the road. While Pheromone which is under way, evaporation tends to be longer. Because ants are under way through more than ants on the way through. 1.d image shows that the other ants in the end decided to skip down the street because there are many Pheromone left. Pheromone while on the road has a lot to evaporate so the ants do not pick on the road. The more ants are under way through the more ants that follow. Likewise, the road, the fewer ants on the way through, then the pheromone left by the reduced and even disappear. From this and elected the shortest route between the nest and the food source Advantages of Ant Colony Algorithm is located on some things that distinguishes this algorithm with the other being:

5 1. The algorithm works with a positive feedback in the discovery and achievement of a good solution. Where at a certain point a group choose a different option and one providing good results, then in the future the choice will always be used. 2. This algorithm has the properties of high synergy. Demonstrated the effectiveness of the search by providing a number of ants working together and each partnership will be mutually independent. 3. Extensive use of structure in the ant algorithm helps in finding an acceptable solution in the early stages of the research process. 4. Splendor of this algorithm to be applied on the same version for the combination of different optimization problems, such as the ATSP ( Asymmetric Traveling Salesman Problem ). 5. This algorithm can be applied to other optimization problems combinations. Stages ant algorithm used in this study are as follows: 1. Prices and parameters initialization phase each ant first node At this stage, the price initialization parameters that will be used next data ant algorithm calculations. This parameter is directly possessed an influence on the calculation of the probability of the node to be visited. Parameters used include a. N is number of city b. X and Y are the coordinates or d is the distance between cities ( but in this paper does not use the X and Y coordinates because it uses the distance d between house to house ) c. Q is a constant cycle of ants d. α is a constant controlling ant trail intensity. e. β is a constant controlling the visibility f. M is the number of ants. g. ρ is the evaporation constant trail of ants h. NCmax is the maximum number of cycles i. τij is the intensity between the ant. 2. Initialize the first node each ant At this stage to determine value of the first node on ants and other ant determines its number 3. Completion of the first node to the taboo list. This stage is node - node tops any all the initialization step 1 should be entered as the first element of the taboo list. Tabu list is used to keep a list of the order of the nodes that have been visited every ant. Then every time an ant visits a city, tabu list element increased by one, and so forth until tabu list is full or has many elements in common with all the nodes that must be visited.

6 4. The preparation of these visits each ant to each node Having ants or the amount distributed to each node, the ants will start the journey from the first node as the origin node and one of its other nodes as the destination. Then the ants will choose one of the next destination node, and so on until all the nodes. So for this purpose code is used to determine the equation named random proportional rule ( Dorigo, M., Maniezzo, V., and Colorni, A., 1996), which is shown by the equation below. the probability of ant k at the point r is chosen to go to the point s.. is Where : (a) s : Pheromone amount contained in the edge between the point and the point r 1 n = : visibility (invers dari jarak d (rs) ) d( rs) (b) ( ( rs) ) (c) α is a parameter that controls the weight ( weight) relative of Pheromone (d) β is the parameter controlling the distance ( α > 0 and β < 0 ) The above equation shows the magnitude of the probability of a node to be selected as the destination node. The greater the price, the greater the probability of the probability to be selected as the destination node. 5. Calculate the length of an ant and the calculation of the intensity of price changes between nodes ant footprint. At this stage of the calculation of the distance between each node. With dij as the distance between the node i to node j is calculated by the equation: This equation is used to determine if known distance using coordinates. But in this study, using the distance between the house so do not use this formula. Ants traveling between nodes will leave footprints at all in its path trajectory. To evaporation leads to changes in the intensity of price footprints between nodes. This changes the equation is : With Dt ij is the price change between the intensity of ant footprints nodes each ant is calculated based on the equation :

7 6. The calculation of the intensity of ant footprints antarnode and reset for the next cycle of price changes intensity antarnode ant footprint.. Ants are used antarnode trajectory travel lane each ant has the possibility to pass the ants in the next cycle. Because of the evaporation of the pheromone and intesitasnya also changed depending on the ants that pass through it in the eyes of ants that will pass on the track for the next cycle, the intensity of the price has changed. Price intensity of inter- node ant footprints for the next cycle is calculated by the equation: Next up for the next cycle of intensity trace all changes in prices between nodes needs to be reset back to zero so valuable. 7. Discharging taboo list If not achieved or the maximum number of cycles has not converged, then the algorithm needs to be repeated again from step 2 to the price of the intensity parameter ant footprints among nodes that have been updated. Besides, taboo list needs to be emptied to be filled again with the new node order in the next cycle. This process goes on until the journey reaches the maximum amount on line NCMax or all ants make the same travel patterns. It is called the stagnation behavior because it is used in situations where the algorithm reaches the point of alternative solutions. 8. The results of data processing using ant algorithm From the data processing using ant algorithm, then obtained the shortest route length of pipe. Can be explained by processing the data as follows: A B C D A B C D

8 Then come to the intensity of the initial trail of ants all over the place, the intensity of the ant trail exist in the following table A B C D A B C D Then the next cycle begins with NC = 1. Where each node is put on all different at the start of the trip. Ant 1 The probability of a node to all the places, the probability obtained by the equation : where is obtained by the equation so that the probability of all to all of the following places : A = 0 B = C = D = cumulative probability= Random Number = then the selected node is B Taboo List : AB Probability of node B to all nodes A = 0 B = 0 C =

9 D = Cumulative Probability = Random number = then the selected node is C Taboo List : ABC Probability of node B to all nodes A = 0 B = 0 C = 0 D = 1 Cumulative Probability = Random number = then the selected node is D Taboo List : ABCD So on can range from ABCD to 55. And in the same way the first ant in obtaining travel route, then the results can be obtained for subsequent ants in the following table Rute Perjalanan Jarak Tempuh Semut 1 ABCD 55 Semut 2 BADC 50 Semut 3 CDBA 40 Semut 4 DABC 45 Then the intensity of the reforms, applying ant trail following equation So that the intensity of the renewal cycle trail of ants on NC = 1 is in the following table A B C D A B C

10 D Calculations will continue to be done to achieve the results NCmax and his grab in which the greatest probability. CONCLUSIONS AND RECOMMENDATIONS Based on the results of the evaluation questionnaire analyst who obtained it can be concluded as follows : 1. Program produced by Ant Colony method as a technique to search can show that the optimal position of the wiring from house to house 2. Ant Colony method used in this program is flexible enough for a search method so that it is possible to evolve according to the existing search. 3. Process of calculating the duration of each algorithm very big influence on the resulting solution. The longer it is used, then the resulting solution ACO algorithm is closer to optimal. 4. The number of iterations required to achieve the best results do not have a positive correlation with the number of points that were completed, but depending on the time required to achieve the best results cabling. Suggestions for the use and application developers: 1. Optimization problem which can be solved by Ant Colony Optimization algorithms in addition to the problem of wiring or cable home much more and this algorithm can be modified according to the application problem solved. Due to the limitations of the author in this final task then there are many things that need to be improved and refined. Therefore, it is expected to continue research on the application of Ant Colony Optimization algorithms that have been modified for the problem - other optimization problems. 2. Ant Colony Optimization in fact there are several types, while the author using Ant Colony System. Therefore, for the future, be able to use Ant Colony Optimization others. REFERENCES Berlianty, Intan., Arifin, Miftahol. Teknik-Teknik Optimasi Heuristik. ISBN : Bullnheimer, B., Hartl, R. F., dan Strauss, C. (1999). An improved ant system algorithm for the vehicle routing problem. Technical report, Institute of Management Science, University of Vienna, Austria. Dorigo, Marco., Colorni, Alberto., Maniezzo, Vittorio., Trubian, Marco (1994). Ant System for Job-shop Scheduling. Dorigo, Marco., dan Gambardella, L. M. (2002). Guest Editorial Special Section on Ant Colony Optimization. Dorigo, M., Maniezzo, V., dan Colorni, A. (1991b). The Ant System: An autocatalytic optimizing process. Technical report revised, Dipartimento di Elettronica, Politecnico di Milano, Milan. Dorigo, M., Maniezzo, V., dan Colorni, A. (1996). The ant system: optimization by a colony of cooperating agents. IEEE Transactions on Systems, Man, and Cybernetics-Part B, vol. 26, no. 2, pp , Dorigo, M., dan Gambardella, L. M. (1997). Ant colonies for the traveling salesman problem. Tech.Rep/IRIDIA/ , Université Libre de Bruxelles, Belgium.

11 Dorigo, M., dan Stu tzle, T. ( 2004). Ant colony optimization. A Bradford book. The MIT Press Cambridge, Massachusetts London, England. Dorigo, M., dan Socha K. (2007), An Introduction to Ant Colony Optimization, Tech.Rep/IRIDIA/ , Université Libre de Bruxelles, Belgium. Edhy Sutanta ( 2004) Entity Relationship Diagram J.L. Bnetley, (1992). Fast algorithms for geometric traveling salesman problems. ORSA Journal on Computing, vol. 4, pp Kenneth E. Kendall, Julie E. Kendall ( 2005) Entity Relationship Diagram Lucas, J.R. dan Henry, C System Analyst and Design. New Jersey: Prentice-Hall. Mutakhiroh, I., Saptono, F., Hasanah, N., dan Wiryadinata, R,. (2007). Pemanfaatan Metode Heuristik Dalam Pencarian Jalur Terpendek Dengan Algoritma Semut dan Algoritma Genetik. Seminar Nasional Aplikasi Teknologi Informasi. ISSN: Yogyakarta. ABOUT AUTHOR Joni Cukri was born in Medan city on 17 November Author completed his undergraduate education at Bina Nusantara University in the field of Computer Science and Mathematics in 2013.