Optimization of Supply Chain Distribution Network Using Cost Matrix Generated by RFID System

1&2

R.Senthilkumar, 2T.Yuvaraj and 3P.Suresh Assistant Professor, 3Professor, Department of Mechanical Engineering Muthayammal Engineering College, Rasipuram-637408. India. E-mail: [email protected]

Abstract- Radio Frequency Identification (RFID) technology has been attracting considerable attention with the expectation of improved supply chain visibility for consumer goods, apparel, and pharmaceutical manufacturers, as well as retailers and government procurement agencies. In this work, an RFID Technique is proposed to Design the Supply Chain Distribution Network system. The RFID system is used to predict the cost matrix of the distribution points of the Supply Chain Network. An improved K-Means hybrid Genetic Algorithm integrated with Simulated Annealing has been proposed to solve the Supply Chain Distribution Network and predicted the optimal output of optimal travel cost. The performance of the proposed system has been compared with existing algorithm and proved this algorithm is produced better results. Keywords: RFID, Genetic Algorithm, Simulated Annealing, Supply Chain Distribution Network Problem, cost matrix, K-Means I. INTRODUCTION

A. Supply Chain Management Supply Chain Management (SCM) is now at the centre stage of manufacturing and service organizations. Supply chain Management is the Procurement of raw material to the shop by the suppliers, from the shop to production unit for manufacturing by the manufacturer, then assembling by the employees and the finished goods to the market by effective way and to customers. The task of managing the entire supply chain constitutes the core of the Supply Chain Management. The type of vehicles and the route to be followed by each vehicle to service customers demands, decisions related with transportation aspect of logistics management are an important part of supply chain decisions. This analysis addresses the transportation of distribution logistics in SCM. The problem has been modelled as a classical routing problem.

B. The Conventional Traveling Salesman Problem The most fundamental and well-studied routing problem is without doubt the TSP, in which a salesman is to visit a set of cities and return to the city he started. The objective for the TSP is to minimize the total distance traveled by the salesman. The VCR is a generalization of the TSP in that the VRP consists in determining m vehicle routes, where a route is a tour that begins at the depot, visits a subset of the customers in a given order and returns to the depot. All customers must be visited exactly once and the total customer demand of a route must not exceed the vehicle capacity. The objective of the VRP is to minimize the overall distribution costs. In most real- life distribution contexts a number of side constraints complicate the model. These side constraints could for instance be time constraints on the total route time and time windows within which the service must begin. The latter problem is referred to ad the VRP with Time Windows (VRPTW). Furthermore, having to deal with aspects such as multiple depots and commodities complicates the models further. Solution methods include exact methods such as mathematical programming, but custom designed heuristics and Meta heuristics such as Genetic Algorithm (GA), tabu search and Simulated Annealing (SA) have also been applied to the VRP. Then, the asymmetric TSP can be formulated as follows: min s.t (1)

CijXij
i =l

Xij = 1
i =l n

where j = 1, , n.,

(2)

Xij = 1 where i = 1, , n.,

where j = 1, , n., i = 1, , n., We set up the problem by numbering each of the n cities as C1,, Cn; a tour is then just a permutation, say (k1, k2, k3,, kn) of (1,, n). We interpret this as the tour which starts at Ck1, then goes to Ck2 and so on, running through the remaining cities up to and including Ckn before finally completing the tour back at Ck1. It is thus clear than we can permute the permutation cyclically without changing the tour. There is no loss of generality then is assuming that Ck1 = C1; it is then clear that there are (n 1)! Different tours. This also shows that solving the problem by exhaustive enumeration is infeasible unless there are very few cities. II. LITERATURE

(3) Xij {0,1}

operator, the model has been transformed into a deterministic multi-objective mixed-integer nonlinear programming model. Then, we use spanning tree-based genetic algorithms by the Prefer number representation to find the SCN to satisfy the demand imposed by customers with minimum total cost and maximum customer services for multiobjective SCN design problem of this company under condition of random fuzzy customers demand and transportation cost between facilities. From the literature, it is observed that the past approaches consider mostly heuristic and very few nontraditional techniques for solving the supply chain distribution network problem. The integration of genetic algorithm with simulated annealing approach has not been given attention in the earlier approaches. III. PROPOSED ALGORITHM Designed dynamic RFID system can generate both distance matrix and cost matrix. In this illustration cost matrix is taken as input data. Algorithm Steps for cost matrix: Step 1: Cost matrix is taken as input (Table) Step 2: Calculate F = N/2 Step 3: Calculate S = (N - (i-1) solution spaces Step 4: Calculate K = N-F Locations Step 5: Initialization and Each solution space will have (N-F)! Sequences. From each solution space K sequences are taken and solution values are averaged to get Average Objective Function Value (AOFV) Step 6: Select the solution space with minimum AOFV. Step 7: OFV is calculated using the below equation:

Chan et al (2007) have proposed a multiple ant colony optimization to design a balanced and efficient supply chain network that maintains the best balance of transit time and customer service. They focused on effective allocation of the customers to the Distributed Centers (DCs) with the two fold objective of minimization of the transit time and degree of imbalance of the DCs. The proposed technique shows better performance considering both negative and positive feedback in search of optimum results. A Mixed Integer Linear Programming (MILP) model have been proposed to describe the optimization problem. A case study for the coatings business unit of a global specialty chemicals manufacturer is used to demonstrate the applicability of the approach in a number of scenarios (Panagiotis Tsiakis et al 2008). Huang et al (2008) have designed a supply chain network in uncertain environment, in which the demands of the customer are taken as random variables and the operation cost involved are programmed using fuzzy neural network and optimized by particle swarm optimization to solve the established model. From the above discussion, it was observed that particle swarm algorithm is successfully applied in various supply chain optimization problems. Jeff Ferrio and John Wassick (2008) have presented a single period network design MILP model for multi-product supply chains. The network is comprised of production plants, an arbitrary number of echelons of distribution centers and customer locations. Jiuping Xu et al (2008) have proposed a random fuzzy multi-objective mixedinteger non-linear programming model for the Supply Chain Network (SCN) design problem, which is representative in the industry of Chinese liquor. By the expected value operator and chance constraint

Min Cost (C) =

f
i =1 j=1

ij

c ij d ij

f = frequency of movement of salesman from i to j c = Transportation cost d = Distance between i to j N = Number of cities 2 3 4 5 6 7

Table. 1. Seven node cost matrix

Cities 1 1 0

44 35 18 28 23 389 1 2 3 4 5

2 3 4 5 6 7

44 1

38 28 27 42 571 1 2 3 4 0 26 14 13 249 1 2 4 0 14 20 621 1 2 0 15 245 7 0 321 0

IV.

RESULTS AND DISCUSSION

35 38 2 1

Table.2. Average Objective Function Value (AOFV) (Initialization process)

S.No

Sequence solution route

AOFV

18 28 26 3 2 1

28 27 14 14 4 3 2 1

23 42 13 20 15 5 4 4 2 7

38 57 24 62 24 32 9 1 9 1 5 1

F = N/2 = 7/2 = 3.5 = Rounded off lower integer value 3. S = (N-1) (N-2) = (7-1) (7-2) = 30 Solution spaces Flow chart for the above algorithm is shown in Figure.1.

1 1- - - - - - - - - - - - 3768.4 2 2- - - - - - - - - - - - 4124.5 3 3- - - - - - - - - - - - 3121.1 4 4- - - - - - - - - - - - 3645.4 5 5- - - - - - - - - - - - 4012.0 6 6- - - - - - - - - - - - 4172.3 7 7- - - - - - - - - - - - 3467.6 In initialization process the Table 2 indicated serial number 3 is shown minimum value. Then Table 3 indicated the average objective function value for all the possible routing.
Table.3. Average Objective Function Value (AOFV)

S.No Sequence solution route 1 2 3 4 5 6 7 8 9

AOFV

S.No Sequence solution route 16 17 18 19 20 21 22 23 24 25 26 27 28 29 3-5-2- - -3-5-1- -- 3-5-6- - 3-5-7- - 3-6-1- - 3-6-4- - 3-6-5- - -3-7-2- - 1-7-4- - --

AOFV

3-2-1- - - - 3342.1 3-2-4- - - - 3124.2 3-2-5- - - - 3651.4 3-2-6- - -3231.4 3-2-7- - - - 3568.4 3-1-2- - - - 3004.5 3-1-4- - 3-1-5- - 3-1-6- - -2342.4 2846.0 2765.3 3457.1 3213.1 3272.4 3072.7

2769.5 3952.6 3233.0 3167.5 3222.6 3022.0 3965.5 2651.6 2891.2

3-5-4- - - - 2948.5

3-6-2- - - - 3121.6

10 3-1-7- - 11 3-4-1- - 12 3-4-2- - 13 1-4-5- - -

3-6-7- - - - 2522.4 3-7-1- - - - 3122.0 1-7-5- - - - 2965.1

14 1-4-6- - - - 3541.4

Figure. 1. Flow chart for the cluster based heuristic algorithm for cost matrix

15 1-4-7- - - - 3210.2 30 1-7-6- - - - 2435.3 1. K= Average Number of Sequences = N-F = 7-3 = 4 2. Number of Sequences = (7-3)! = 24 3. Optimal Sequence = 7th Item = 3-1-4- - - - - - - (Generate all 24 possible generations) 4. Best optimal route = 3-1-4-6-5-7-2 = Minimum OFV = 2297.1. 5. The proposed algorithm gives the better results when compared with the optimal layout sequence of 4-6-5-1-7-2-3 obtained by Lee et al (2001) with an OFV of 3198. In this numerical illustration, the performance and computational efficiency of the proposed algorithm

are analyzed. Also a discussion on the factors influencing the results of the algorithm is provided. V. VALEDICTION OF PROPOSED ALGORITHM

The validity of the proposed model is tested with the realistic problems taken from Reinelt (1994). The results shows (Table 4) the proposed algorithm is produced better results and nearer to the solutions given in TSPLIB95.
Table.4. Valediction of Existing problems using proposed algorithm

[1] Altiparmak F., Gen M., Lin L. and Paksoy T. (2006), A genetic algorithm for multi-objective optimization of supply chain networks, Computers and Industrial Engineering, Vol.51, pp.197-216. [2] Chan F.T.S., Kumar N. and Choy K.L. (2007), Decision making approach for the distribution centre location problem in a supply chain network using the fuzzybased hierarchical concept, Proceedings of The Institution of Mechanical Engineers Part B: Journal of Engineering Manufacture, Vol.221, pp.725-739.

[3]

Chiung Moon et al., (2002) An efficient GA for the TSP with precedence constraints. European Journal of Operational Research. 140, 606-617. [4] Dantzing, B.G. and J.H. Ramser (1959) The truckdispatching problem Management Sciences, 6(1), 80-91.

[5] [6]

Giovanni Righini and Marco Trubian (2002) A note on the approximation of the asymmetric TSP. Discrete Optimization, 13 May. In Chan Choi et al., (2003) A GA with a mixed region search for the asymmetric TSP. computers & Operations Research 30, 773-786. [7] Jeff Ferrio and John Wassick (2008), Chemical supply chain network optimization, Computers and Chemical Engineering, Vol.32, pp. 2481-2504. [8] Jiuping Xu, Qiang Liu and Rui Wang (2008), A class of multi-objective supply chain networks optimal model under random fuzzy environment and its application to the industry of Chinese liquor, Information Sciences, Vol.178, pp.2022-2043.

Problem
Burma14 (TSP-sy) BR17 (TSPunsy) ULYS22 (TSP-sy) BAYS29 (TSP-sy)

Shortest Distance
30.87 40.09 77.84 2125

Optimal Route
8-13-7-12-6-5-4-3-14-2-110-9-11-8 2-14-3-12-1-16-15-7-6-5-417-9-8-13-11-10-2

GASA
31.32 38.20

14-12-13-3-2-17-18-4-22-8- 73.27 1-16-15-5-11-9-10-21-20-197-6-14 23-8-24-27-16-19-13-1-28-6- 2012 12-9-5-26-29-3-2-21-20-104-15-17-18-14-22-11-25-723

[9]

Kirkpatrick, S., C. D. Gelatt, Jr. and M.P. Vecchi (1983) Optimization by SA, Science, Number 4598, 13 May.

In all the problem are assumed and 100 iterations are performed and the results are given and compared with exact results. The matrix in all the problems given show the distances between the cities i to j. The computer time for smaller size problems say less than 20 cities only below two minutes. All the programs are executed on Pentium IV processor under Microsoft windows 2003 operating system. VI. CONCLUSION

[10] [11] [12] [13]

Lenstra, J. and K.A. Rinnooy (1981) Complexity of vehicle routing and scheduling problems. Networks, 11 221-227. Narahari, Y and S. Biswas (2000) supply chain management the role of models proceedings of the international conference on intelligent flexible autonomous manufacturing systems, January 10-12. Panagiotis Tsiakis and Papageorgiou L.G. (2008), Optimal production allocation and distribution supply chain networks, International Journal of Production Economics, Vol.111, pp.468-483. Reinelt G. (1994), The Traveling Salesman: Computational Solutions for TSP Applications, LNCS, Springer Verlag, Vol.840. [14]Srivastava S.K. (2008), Network design for reverse logistics, Omega, Vol.36, No.4, pp.535-548. [15]Steuer R.E. (1986), Multiple Criteria Optimization: Theory, Computation and Application, Wiley, New York. [16][email protected] and [email protected] [17]www.chipdocs.com/datasheets/datasheet-pdf [18]www.logistik-heute.de [19]Zegordin S.H., Itoh K. and Enkawa T. (1995), Minimizing makespan for flow shop scheduling by combining simulated annealing with sequencing knowledge, European Journal of Operations Research, Vol.85, pp.515-531. [20]Zhou G., Min H. and Gen M. (2002), The balanced allocation of customers to multiple distribution centers in the supply chain network: A genetic algorithm approach, Computers and Industrial Engineering, Vol.43, pp.252-261.

Radio Frequency Identification (RFID) technology has been attracting considerable attention with the expectation of improved supply chain visibility for consumer goods, apparel, and pharmaceutical manufacturers, as well as retailers and government procurement agencies. In this work, an RFID Technique has been proposed to Design the Supply Chain Distribution Network system. The RFID system is used to predict the cost matrix of the distribution points of the Supply Chain Network. An improved K-Means hybrid Genetic Algorithm integrated with Simulated Annealing has been proposed to solve the Supply Chain Distribution Network and predicted the optimal output of optimal travel cost. The performance of the proposed system has been compared with existing algorithm and proved this algorithm is produced better results. REFERENCES