Related Works Theoretical Review

101 ISSN 2086-5953 ANT COLONY ALGORITHM BASED GENERIC MODEL FOR SUPPLY CHAIN PERFORMANCE MEASUREMENT Taufik Djatna 1 , Ditdit N. Utama 2 1 Postgraduate Program of Agro-Industrial Technology, Graduate School of Bogor Agricultural University, Darmaga P.O. Box 220 Bogor 16002, Indonesia 2 Information System Department, Science and Technology Faculty of State Islamic University Syarif Hidayatullah – Jakarta, Ir. H. Juanda Street No. 95 Ciputat, Jakarta, Indonesia E-Mail: taufikdjatnaipb.ac.id 1 , ditditnhotmail.com 2 ABSTRACT Ant colony algorithm is one of many mature research methodologies in searching, routing and optimizing field. This methodology uses biological and behavior phenomenon as an algorithm philosophy to solve fastest path problem and other optimization problems. Many researchers have researched by using ant colony algorithm as their methodology, not only for technical cases, but also for management aspect cases. The ant colony algorithm methodology can be used in problem solving for the routing or optimizing supply chain path in supply chain management case. As we know, that in the supply chain case, the best path searching problem is a usual problem happen. In this paper, the authors propose a generic model based on ant colony algorithm that was implemented in supply chain performance measurement cases. Keywords: ant colony algorithm, generic model, supply chain, performance measurement 1 INTRODUCTION 1.1 Background More than one company involvement in industrial sector is the reason supply chain management concept is needed in industrial implementation. Relationship among companies in one industrial area, such as: raw material producers, collectors, factories, warehouses, distributors, retailers, and the customers, is a supply chain flow that could not be separated one to each others. The differentiation of data and information platform, leadership style, and the working culture in every level of supply chain, is the problem that has to be solved. Supply chain is a system where the organizations or companies can supply their products and services to their customers. This chain is as network of many related companies that have the same goals and objectives. The supply term, in supply chain term, includes product transformation, in this case, transforming raw material to end product [1]. On the other hand, the best path of supply chain searching problem, to supply the product from supply to demand side [2], always be researched in day per day by many researchers. In other research, [3] has researched in supply chain management as Intelligent Supply Chain. ISN is a system that integrates all of the involved actors in supply chain. [4] has researched in responsive and efficient supply chain. One of many methodologies that can be used for solving the current supply chain problem is ant colony algorithm methodology, or usually called Ant Colony Optimization ACO. ACO is a paradigm for designing meta-heuristic algorithms for combinatorial optimization problems. The essential trait of ACO algorithms is the combination of a priori information about the structure of a promising solution with a posteriori information about the structure of previously obtained good solutions [5]. Other definition said that ACO is a heuristic optimization method for the fastest path problem solving or other problem optimization by using biological ant behavior phenomenon [6]. The meaning of generic model in this case is the model that describes the developed model generally. On the other hand, the generic model is the opened model. It can be extended in accordance with what we want, which means our model will support solution in different process in the supply chain. According to that background, this paper would explained research background, related works, theoretical review in the first part; generic model development in the second part; measurement and result in the third part; and discussion and conclusion in the fourth part.

1.2 Related Works

There are many researches that used ant ISSN 2086-5953 colony optimization as a main technique. Some of them are [7] researched in production optimization and maintenance process scheduling area; [8] researched by using ACO in routing design area; [9] have researched in project critical path searching area by using ACO as a main method [10] also studied in product packing and delivering area for supply chain; and [11] have studied aspect of supply chain in performance management, in combination of SCOR, value added, cost and financial supply chain performance measurement. On the other hand, [12] combined ACO and Genetic Algorithm for developing the Chinese question and answering system. [13] used ACO for data processing dynamic city choice case and [14] used ACO for partner selection case. Moreover, [15] and [16] used ACO that was implemented in mobile system case.

1.3 Theoretical Review

Supply Chain Management Supply chain concept is still a growing concept in logistic problem view. In a conventional concept, logistic problem has been seen as internal company problem only. The companies have to solve their problems internally and solely, without collaborate to their company partners. In this new one, logistic problem is seen as a boarder problem. The logistic problem can be happen in every step of supply chain, from the first level up to the last level of supply chain. That is reason why supply chain management can be defined as a set of approaches utilized to efficiently integrate suppliers, manufactures, warehouse, and stores, so that merchandise is produced and distributed at the right quantities, to the right locations, at the right time, in order to minimize system wide costs while satisfying service level requirement [2]. The figure 1 described supply chain management concept [2]. Figure 1. Concept of Supply Chain Management On the other hand, supply chain can be defined as a system that used by company to supply its product and service to its customers. This chain is also as network of many related organizations that have the same goals and objectives [1]. Ant Colony Optimization [17] Said that ACO is an optimization technique inspired by observations on the behavior of biological ant colonies; it has been introduced by [13], and [19]; and developed later into a meta- heuristic for combinatorial optimization problems [20], [21], [22]. Recently, ACO algorithms have also turned out as competitive in a discrete stochastic optimization context [23], [24], [25], [26], [27]. An ACO is essentially a system based on agent, which simulates the natural behavior of ants including mechanisms of co-operation and adaptation. To use this kind of system, a new meta- heuristic is proposed in order to solve combinatorial optimization problems. This new meta-heuristic has been shown to be both robust and versatile in the sense that it has been successfully applied to arrange of different combinatorial optimization problems. ACO algorithms are based on following ideas [28]: 1. Each path, is followed by an ant, is associated with a candidate solution for a given problem. 2. When an ant follows a path, the amount of pheromone deposited on that path. This path is proportional to the quality of the corresponding candidate solution for the target problem. 3. When an ant has to choose between two and more paths, the path with a larger amount of pheromone has a greater probability of being chosen by the ant. Conceptually, in ant colony algorithm, two ways of food searching way, that are used by ants, are used in this method. Other principles of ACO are [29]: 1. Random direction for searching and collecting some food, 2. When one ant finds food resource, that ant traces back again the path that was used by it, 3. The ants put chemical Pheromones on their path that can be evaporated, 4. In loop, the chemical pheromones will decreases, but on the other hand, it will be added, 5. Some ants will follow the shortest path, the strongest chemical pheromone path. 2 GENERIC MODEL Based on basic behavior of ACO model [22], we generate a generic mathematical model for supply chain performance measurement which was described in the next formula. ISSN 2086-5953 Where: = Performance of Current Path = Current Distance, the basis of measurement = Performance of Previous Node Variable = Total of Previous Node Variable = Current Path Pheromone The supply chain performance depends on the path length in this case, distance is the basis variable for measurement, supply chain element performance and pheromone quality. The supply chain element performance could be showed in the next mathematic formula. Where: = Performance of previous element = Performance of prev ious element‘s SCOR variable = Performance of Previous element‘s Financial variable = Performance of Previous element‘s Added Value Variable = Performance of previous element‘s Transportation Cost Variable The supply chain element performance depends on performance of SCOR, financial, added-value and transportation cost variables. This is an opened model, so if it is needed, we can add some new variables, according to the case that was being faced. The generic model can be described in usecase diagram. This model is divided in to two sub-systems; Ant colony sub-system and Supply Chain Performance Sub-System. In the ant colony sub-system, user can use the main model to search the best path of supply chain. On the other hand, supply chain and supply chain element performance are measured in supply chain performance sub- system. The usecase diagram could be showed in Figure 2. Information flow and type of the model could be described by using information liquid model Figure 3. This model could be showed some information resources that were needed in this model, such as: supply chain element; value of SCOR Based on [30], financial, added value and transportation cost variable; and distance and route. The generic model is shown in class diagram Figure 4. To describe the model, there are five classes in this class diagram: class AntGraph, class Ant, class AntColony, class Ant4SC and class AntColony4SC. Class AntGraph connects to ClassSupplyChainPerformance that consists of: class SCORPerformance, class financialPerformance, costPerformance, and addedValuePerformance. Each class represents the name of main component in ACO. This model can be called by distance based model [11]. Ant Colony Sub-System Supply Chain Performance Sub-System User The Best SC Searching Criteria Identification SCOR Performance Added Value Variable Performance Financial Variable Performance SC Element Performance SC Performance Transportation Cost Variable Performance Figure 2. Usecase Diagram of Generic Model Figure 3. Information Liquid Model ISSN 2086-5953 0..1 0.. 0..1 0.. antGraph 0..1 0.. 0..1 0.. 0..1 0.. 0..1 0.. 0..1 0.. 0..1 0.. 0..1 0.. 0..1 0.. ant - - - - - - - - - - - m_nAntID m_path m_nCurNode m_nStartNode m_dPathValue m_pathVect s_nAntIDCounter s_dBestPathValue s_bestPathVect s_bestPath s_nLastBestPathIteration : int : int[][] : int : int : double : int : int : double : int : int[][] : int + + + + + + + + + + setAntColony reset ant int nStartNode, Observer observer init start run better double dPathValue, double dBestPathValue stateTransitionRule int r localUpdatingRule int r, int s end getBestPath : void : void : void : void : void : boolean : int : void : boolean : int[] antColony - - - - - m_ants m_nAnts m_nAntCounter m_nIterCounter Attribute_5 : int : int : int : int : int + + + + + + + + + + + + + antColony int nAnts, int nIterations createAnts int ants getBestPathVector start iteration getGraph getAnts getIterations getIterationCounter getID update Obeservable ant, Object obj getBestPathValue getBestPath getLastBestPathIteration done globalUpdatingRule : Ant[] : vector : int : void : int : int : int : int : void : double : int[] : int : boolean : void AntGraph - - - - m_delta m_tau m_nNodes m_nTau : double[][] : double[][] : int : double + + + + + + + + + + antGraph int nNodes, double[] delta[], double[] tau[] delta int r, int s tau int r, int s etha int r, int s nodes tauD updateTau int r, int s, double value averageDelta averageTau average double[] matrix[] : double : double : double : int : double : void : double : double : double supplyChainPerformance - - IDPerformance SCPerformance : String : double + + dataLoad supplyChainPerformanceMeasurement : double : double Distance - - IDDistance distance : String : double + + distanceMeasurement distanceLoad : double : double SCORPerformance - - IDSCOR SCORValue : String : double + + SCORValueMeasurement dataView : double : void financialPerformance - - IDFinancial financialPerformanceValue : String : double + + financialPerformanceMeasurement dataView : double : void costPerformance - - IDCost CostPerformanceValue : String : double + + costPerformanceMeasurement dataView : double : void addedValuePerformance - - IDAddedValue addedValuePerformance : string : double + + addedValuePerformanceMeasurement dataVeiw : void : void DistPar - - Distance ParentVert : double : int + DistPar : void Vertex - - Label isInTree : char : boolean + Vertex : void Figure 4. Class Diagram of Model The detailed ant colony based generic model could be described in Figure 5. It is started at variable initiations, ants generating, up to supply chain performance value getting. Further explanation of this model could be described in sequence diagram in Figure 6. This diagram explained the relationship among objects or classes in the model. In the other words, this diagram explained the relationship among entities in the model. SequenceDiagram_1 PoCost PoAddedValue PoFinancial PoSCOR SumOfAnt pheromones, Evaporation OptimumPath call call call call graph call call graph graph generate Actor_1 Graph Ant AntColony DistPar Vertex SupplyChainPerformance SCORPerformance FinancialPerformance CostPerformance AddedValuePerformance PoCost PoAddedValue PoFinancial PoSCOR SumOfAnt pheromones, Evaporation OptimumPath call call call call graph call call graph graph generate Figure 6. Sequence Diagram of Model Figure 5. Activity Diagram of Model ISSN 2086-5953 3 MEASUREMENT AND RESULT In order to justify our model, we do some measurement based on Figure 4. Data and information resource for this step is derived from field observation and computation. We verify the ACO generic model on an example for this measurement which is used for supply chain performance measurement case was described in a tree view in Figure 7. We deploy a spanning tree algorithm to construct this tree. The authors take this example from palm based bio-energy supply chain case. The supply chain network described that A is the end customer; B and C are bio-energy factories; D is a Crude Palm Oil CPO factory; and E, F and G are palm farmers. The distance value of the tree was converted in to graph matrix in Table 1. This matrix described distance between nodes or elements in one supply chain network. A B C D F E G 55 75 40 55 50 45 35 Figure 7. Supply Chain Tree Table 1. Graph Matrix A B C D E F G A 0.00 55.0 75.00 0.00 0.00 0.00 0.00 B 0.00 0.00 0.00 40.0 0.00 0.00 0.00 C 0.00 0.00 0.00 55.00 0.00 0.00 0.00 D 0.00 0.00 0.00 0.00 50.0 45.00 35.0 E 0.00 0.00 0.00 0.00 0.00 0.00 0.00 F 0.00 0.00 0.00 0.00 0.00 0.00 0.00 G 0.00 0.00 0.00 0.00 0.00 0.00 0.00 The performance value of each element: SCOR, financial, cost and added value, could be shown by data in Table 2. Table 2. Performance Value A B C D E F G SCOR 0.7 0.7 0.5 0,5 0.5 0.5 0.7 Financial 0.5 0.7 0.7 0.7 1.0 0.5 0.7 Cost 0.7 0.7 0.5 0.7 1.0 0.5 0.7 Added Value 0.7 0.7 0.5 1.0 0.5 0.5 0.5 The performance evaluation could be explained by using result of the model execution. Some data resulted from this execution are in Table 3. Table 3. Execution Result ∑Ph. ∑Ev. Path Distance Performance Value 0.11 0.01 A,C,D,F 175 50.77 0.13 0.03 A,C,D,E 180 29.92 0.01 0.15 A,B,D,E 145 81.45 0.05 0.05 A,B,D,E 145 61.13 0.05 0.07 A,B,D,E 145 73.76 ∑Ph. = Average of Pheromones, ∑Ev. = Average of Evaporation, Sum of Ant = 2000 The execution result showed that optimum path is A, B, D, E and the best performance value is 61.13, the most little one. On the other hand, the best performance value is 29.92 with path: A, C, D, E.This performance value was described in km kilometers unit size. 4 CONCLUSION AND DISCUSSION Ant Colony Algorithm could be implemented in model of supply chain performance measurement. It was called ant colony algorithm based generic model for supply chain performance measurement. Usecase diagram showed the relationship between user and the model, and between both sub- systems; ant colony and supply chain performance sub-system. This generic model is an opened model, because there are many variables that can be added in this model, it depends on the case was being faced. In this paper case, author used 4 variables: SCOR, financial, added value and transportation cost variable. REFERENCES [1] Indrajit RE, and Djokopranoto R. 2002. Supply Chain Management Concept. Grasindo [2] Levi DS, Kaminsky P, and Levi ES. 2000. Designing and Managing the Supply Chain: Concepts, Strategies and Case Studies . Irwin McGraw-Hill [3] Rajamani D. 2005. Intelligent Supply Network . The University of Texas School of Management [4] Hugos M. 2006. Essentials of Supply Chain Management . John Wiley Sons, Inc. Second Edition [5] Maniezzo V, Gambardella LM, and Luigi FD. 2004. Ant Colony Optimization. Internet Publication [6] Habilation LZ, and Hammer B. 2003. Ant Colony Optimization . Osnabruck University ISSN 2086-5953 [7] Berrichi A, Yalaoui F, Amodeo L, and Mezghiche M. 2010. Bi-Objective Ant Colony Optimization Approach to Optimize Production and Maintenance Scheduling. Computers and Operations Research 37 2010: 1584 – 1596 [8] Barcos L, Rodriguez V, Alvarez MJ, and Robuste F. 2010. Routing Design for Less- Than-Truckload Motor Carriers Using Ant Colony Optimization. Transportation Research Part E 46 2010: 367 – 383 [9] Duan Q, and Liao TW. 2010. Improved Ant Colony Optimization Algorithms for Determining Project Critical Paths. Journal of Automation in Construction 2010: 1 – 18 [10] Catay B. 2010. A New Saving-Based Ant Algorithm for the Vehicle Routing Problem with Simultaneous Pickup and Delivery. Journal of Expert System with Applications 2010: 1 – 9 [11] Djatna T, and Utama DN. 2010. Modeling UML for Optimization of Supply Chain in Palm Oil Based Bioenergy. International Conference – AFITA 2010 [12] Di S, He P, and Li H. 2010. Research and Implement on Genetic Algorithm and Ant Colony Algorithm in Chinese Question and Answering System. Proc. Second Int Computer Engineering and Applications ICCEA Conf, 2010, 1, 166-169 [13] Yun BJ, and Yong LS. 2009. Improved Ant Colony Algorithm with Emphasis on Data Processing and Dynamic City Choice. Proc. Int. Conf. Information Engineering and Computer Science ICIECS 2009, 2009, 1-4 [14] Wei DH. 2009. Improved Ant Colony Algorithm for Partner Selection. Proc. Int. Conf. Management Science and Engineering ICMSE 2009, 2009, 265-270 [15] Wang JZ, and Hong W. 2009. Application of an Ant Colony Algorithm Based on Complex Networks in Migration of Mobile Agents. Proc. IEEE Int. Symp. IT in Medicine Education ITIME 09, 2009, 1, 821-825 [16] Jin WC, Jie WW, and Hong J. 2009. Cross Layer Ant Colony Algorithm for Cooperative Relay Multicast Tree in Cellular Mobile System. Proc. Fifth Int. Conf. Natural Computation ICNC 09, 2009, 3, 30- 34 [17] Gutjahr WJ. 2006. On the Finite-Time Dynamics of Ant Colony Optimization. Methodol Comput Appl Probab 2006 8: 105 –133 [18] Dorigo M, Maniezzo V, and Colorni A. 1991. The Ant System: an autocatalytic optimization process. Technical Report 91- 016, Dept. of Electronics, Politecnico di Milano: Italy [19] Dorigo M, Maniezzo V, and Colorni A. 1996. Ant System: optimization by a colony of cooperating agents. IEEETransactionson Systems,Man, and Cybernetics, Part B vol. 261: 29 –41 [20] Dorigo M and Di Caro G. 1999. The Ant Colony Optimization metaheuristic, In D. Corne M. Dorigo, and Glover F, eds., New Ideas in Optimization, 11 –32, McGraw Hill: London, UK [21] Dorigo M, Di Caro G, and Gambardella LM. 1999. Ant algorithms for discrete optimization, Artificial Life vol. 52: 137 – 172 [22] Dorigo M and Stutzle T. 2004. Ant Colony Optimization, MIT Press: Cambridge, MA [23] Bianchi L, Gambardella LM, and Dorigo M. 2002. Solving the homogeneous probabilistic travelling salesman problem by the ACO metaheuristic. Proceedingd ANTS ‘02, 3rd International Workshop on Ant Algorithms: 177 –187 [24] Gutjahr WJ. 2003. A converging ACO algorithm for stochastic combinatorial optimization. Proceedings SAGA 2003 Stochastic Algorithms: Foundations and Applications, In A. Albrecht and K. Steinhoefl eds., Lecture Notes in Computer Science 2827, 10 –25, Springer: Berlin, Germany [25] Gutjahr WJ. 2004. S-ACO: an ant-based approach to combinatorial optimization under uncertainty, Ant Colony Optimization and Swarm Intelligence: Proceedings of ANTS 2004 —Fourth International Workshop, Lecture Notes in Computer Science 3172: 238 –249, Springer: Berlin, Germany [26] Rauner M, Brailsford SC, Gutjahr WJ, and Zeppelzauer W. 2005. Optimal screening policies for diabetic retinopathy using a combined discrete-event simulation and ant colony optimization approach. In J. G. Anderson and M. Katzper eds., Proceeding International Conference on Health Sciences Simulation, Western MultiConference ‘05: 147 –152 [27] Birattari M, Balaprakash P, and Dorigo M. 2005. ACO F-Race: Ant colony optimization and racing techniques for combinatorial optimization under ISSN 2086-5953 uncertainty. Proceedings of MIC 2005 —6 th Metaheuristics International Conference: 107 –112 [28] Sinha AN, Das N, and Sahoo G. 2007. Ant Colony Based Hybrid Optimization for Data Clustering. Kybernetes Vol. 36 No. 2, 2007: 175-191 [29] Ahmad MA, and Srivastava J. 2008. Expert Identification in Social Networks an Ant Colony Optimization Approach . University of Minnesota [30] Bolstorff P, and Rosenbaum R. 2007. Supply Chain Excellence - A Handbook for Dramatic Improvement Using the SCOR Model . Amacom. ISSN 2086-5953 [This page is intentionally left blank] 109 ISSN 2086-5953 ARTIFICIAL NEURAL NETWORK TO MEASURE PHOSPHORUS AMOUNT IN AGRICULTURAL SOILS Feriyonika 1 , Nasih Widya Yuwono 2 1 Graduate School of Automation and Control, National Taiwan University of Science Technology 2 Soil Science Department, Faculty of Agriculture, Gadjah Mada University Email: feriyonikayahoo.com 1 , nasihugm.ac.id 2 ABSTRACT This research investigates alternative way to measure one of important elements, phosphorus, in agricultural soils. In many agricultural lands, the farmers‘ behavior is just using their experience and invalid information when take fertilizer in their soil. This condition not only scarifies the soil fertility and harm microorganism but also waste the farmer‘s money. This study develops equipment using camera, computer and Artificial Neural Network ANN as algorithm. The result show that proposed method can do measure the phosphorus amount well. Keywords: ANN, agriculture, soils, camera, phosphorus. 1 INTRODUCTION The Soil holds important role for the plant growth. It has main function as place to growth, providing water and providing air to root breath. There are three main factors deciding the quality of fertility in agricultural soil. They are Biology, physic and chemistry factors. In real world, they influence one to another and must be equilibrium [1]. If the plants can get the nutrition directly from the soil, they will grow well and result good crop. Today, the nutrition for the plants provided by the agricultural soil is not enough. Since the industrial Green Revolution of the 20th century, chemically synthesized inorganic fertilizers were widely used in agricultural plant. It replaces organic fertilizer due to fulfill plant nutrition fast. Lately, it has been realized that the utilization of the chemically synthesized inorganic fertilizers causes decreasing of microorganisms in the soil and affect the decreasing of the soil fertility. In Indonesia, a country with wide agriculture land, the utilization of the chemically synthesized inorganic fertilizers has started since early 70-es. About twenty years later, early 90s, the crop has not satisfied anymore. The condition of soil is very important information for the farmer before they cultivate the plant. The information will be useful to decide what the elements being needed are and how many is. This procedure is not work well due to difficulties to conduct it. The farmers, who leave far from the center of research or university, do the fertilizing based on experience and lack of information. This condition causes disadvantage for the farmers; wasting fertilizer and money. In this research, we use artificial neural network ANN to predict one of important elements, phosphorus P, deciding the quality of soil‘s fertility. There have been researches in agricultural field involving soft computing intelligence techniques. He et al. [2] use ANN to estimate monthly total nitrogen concentration in streams. They investigate the relationship among the land use, fertilizer, and hydrometerological conditions in 59 river basins over Japan and then applied to estimate the monthly river total nitrogen concentration TNC. Anctill et al. [3] use ANN to simulate daily nitrate-nitrogen and suspended sediment fluxes from a small agricultural catchment Melarches, 70 km east of Paris, France. They investigate the soil erosion and biochemical applications associated to human activities such as intensive agriculture. Lek et al. [4] used ANN to predict phosphorus concentration and phosphorus load from watershed characteristics. They investigate relationship between land use and the concentration and export of total phosphorus and of orthophosphate by watersheds. Salehi et al. [5] used ANN to predict of annual nitrate-N N0 3 -N losses into drain flow at Eugene F. Whelan Experimental Farm Agriculture Canada, Woodslee, Ontario, Canada. They used daily measurements of nitrate- N taken from eight different soil conservation treatments during 1992-1994. In this research, we use ANN to predict phosphorus amount in agricultural soils. The result of this research is aim to make portable instrument included ANN inside which can be used in many place easily. The instrument is helpful tool to give farmer information about soil condition. The information is very important because farmer can give appropriate treatment to their agricultural land. ISSN 2086-5953 2 MODEL, ANALYSIS, DESIGN, AND IMPLEMENTATION

2.1 Proposed method