International Journal of Computational Intelligence Systems

Volume 11, Issue 1, 2018, Pages 195 - 205

Discrete Firefly Algorithm for Clustered Multi-Temperature Joint Distribution with Fuzzy Travel Times

Authors
Sichao Lu1, *, lusichao@163.com, Xifu Wang2, wxfky@vip.sina.com
1School of Traffic and Transportation, Beijing Jiaotong University, Beijing, 100044, China
2School of Traffic and Transportation, Beijing Jiaotong University, Beijing, 100044, China
*Corresponding author.
Corresponding Author
Received 9 February 2017, Accepted 15 October 2017, Available Online 1 January 2018.
DOI
10.2991/ijcis.11.1.15How to use a DOI?
Keywords
MTJD, distribution; Z-shaped; Firefly algorithm; Perishable food
Abstract

This study proposes a mathematical model of clustered multi-temperature joint distribution in fuzzy environment. In this model, a Z-shaped function is used to depict customer satisfaction. For the imprecise model, triangular fuzzy numbers are used to represent travel times. By redefining the movement procedure of fireflies, two versions of discrete firefly algorithms are developed, which differ in the population initialization strategy. Finally, experiments are carried out and computational results are reported to confirm the effectiveness of the proposed algorithms.

Copyright
© 2018, the Authors. Published by Atlantis Press.
Open Access
This is an open access article under the CC BY-NC license (http://creativecommons.org/licences/by-nc/4.0/).

1. Introduction

Nowadays, many logistic enterprises have configured the food cold chain to maintain food quality, because refrigeration can reduce the deterioration rate of perishable food products. Consequently, optimization of perishable food delivery has been widely studied by academics and practitioners in the catering industry and the logistics industry. A comprehensive review of the literature shows that, many mathematical models which deal with this problem are based on the model of the vehicle routing problem (VRP)[16]. However, these models are constructed on the assumption that perishable food products with different temperature demands can be put together in a vehicle with a single temperature zone. Therefore, they are not applicable to the route optimization of multi-temperature joint distribution (MTJD), which is a novel approach to the distribution of perishable food products.

In recent years, due to rapid advances in e-business and web technologies, perishable food products such as vegetables and ice creams can be ordered and purchased conveniently through the Internet. As a result, the cold distribution market and the home delivery market have undergone a rapid expansion, which contributes to the development of multi-temperature joint distribution (MTJD). The MTJD approach can be divided into two types[7]. The first type is the mechanical refrigerated compartment division. This uses a technique which can divide a single vehicle compartment into different temperature zones. In the second type, standardized cold insulated boxes and cabinets are placed in regular vehicles to store perishable food products at varying temperatures. The MTJD system is able to reduce logistics costs significantly and attain customer satisfaction, as well as guaranteeing food safety and food quality[8]. However, only a few papers refer to route optimization and load planning in MTJD. Cho and Li[9] formulated the mathematical model of the multi-temperature refrigerated container vehicle routing problem and proposed a two-stage heuristic algorithm. Hsu et al.[10] modelled facilities planning for MTJD in a hierarchical hub-and-spoke network. Sun[11] proposed a bi-objective MTJD which aims at minimizing cost and total carbon emissions, and solved it by using the augmented ε-constraint method. Lu and Wang[12] put forward two suggestions for extending the mathematical model of MTJD to a further step. First, travel times can be defined as fuzzy numbers. Second, a Z-shaped curve can be used to depict the relationship between customer satisfaction and arrival time, rather than linear curves.

It is clear that MTJD is highly combinatorial. Consequently, exact algorithms which aim at obtaining a global optimal value may be far less efficient than heuristic algorithms for most medium-sized and large-sized problem instances. Hence, this paper attempts to develop an evolving algorithm to obtain near-optimal solutions. As a popular evolutionary algorithm which is inspired by the collective behaviour of fireflies, the firefly algorithm (FA) is applicable to almost all engineering areas[13]. This algorithm was first proposed by Yang, who claimed it was superior to genetic algorithms and particle swarm optimization[14]. Given that the original FA was designed to solve continuous optimization problems, some researchers have proposed new versions of the firefly algorithm to deal with discrete optimization problems such as the two-dimensional min-cost covering problem[15], the asymmetric and clustered vehicle routing problem with simultaneous pickups and deliveries[16] and the fuzzy cold storage problem[17].

This paper presents a study of a clustered MTJD, which uses cabins in conventional vehicles to distribute perishable food products. Section 2 gives the problem definition and the mathematical formulation of clustered MTJD. Section 3 presents the discrete firefly algorithm-based solution approaches in detail. In section 4, the computational performances of the proposed algorithms are evaluated, followed by a conclusion in section 5.

2. Mathematical Model

2.1. Fuzzy preliminaries

Since the clustered MTJD involves fuzziness, various fuzzy concepts are discussed briefly in this subsection.

2.1.1 LR-type fuzzy number

A fuzzy number M˜=(m,α,β)LR is called LR-type if there exist decreasing functions L (for left), R (for right), and real numbers α > 0, β > 0 with m, whose membership function can be characterized by Eq. (1)[18].

μM˜(x)={L(mxα)xmR(xmβ)xm

M˜ is called positive if μM˜(x)=0, ∀ x < 0. Based on Eq.(1), an LR-type trapezoidal fuzzy number T˜=(m1,m2,α,β)LR can be defined as[19]:

μT˜(x)={L(m1xα)xm11m1xm2R(xm2β)xmm2

2.1.2 Triangular fuzzy number

A triangular fuzzy number à can be defined by a triplet (al, ac, ar), where the membership function can be determined by:

μA˜(x)={0x<alL(x)=1acxacalalxacR(x)=1xacaracacxar0ar<x

2.1.3 Graded mean integration representation

The graded mean integration representation (GMIR) method is a popular defuzzification technique and has many applications, such as defuzzifying the parameters of an economic order quantity model[20] and those of the shortest path problem[21]. Let L−1 and R−1 be the inverse functions of the functions L(x) and R(x), respectively. The graded mean h-level value of the LR-type trapezoidal fuzzy number T˜=(m1,m2,α,β)LR is h[L−1(h)+R−1(h)]/2 as in Fig. 1. Then the GMIR of T˜ is determined by[22]:

Fig. 1.

The graded mean h-level value of T˜=(m1,m2,α,β).

P(T˜)=01h[L1(h)+R1(h)]/2dh/01hdh

The positive triangular fuzzy number à = (al, ac, ar) can be regarded as an LR-type trapezoidal fuzzy number à = (ac, ac, ac-al, ar-ac)LR. Thus, L−1(x) and R−1(x) of à are given in Eq. (5) and Eq. (6) respectively.

L1(x)=(acal)x+al
R1(x)=(acar)x+ar
Based on Eq. (4)Eq. (6), the GMIR of à can be acquired by using Eq. (7)
P(A˜)=01h[L1(h)+R1(h)]/2dh/01hdh=01h[(acal)h+al+(acar)h+ar]/2dh/01hdh=01(ach2alh2+alh+arh+ach2arh2)/2dh/01hdh=12(ac3al3+al2+ar2+ac3ar3)/12=al+4ac+ar6

2.1.4 Representation of operations between triangular fuzzy number and a crisp number

Let à = (al, ac, ar) and B˜=(bl,bc,br) be two triangular fuzzy numbers and let k be a positive crisp number. According to the fuzzy arithmetical operations under the function principle, the addition operation ⊕, the subtraction operation ⊖, the multiplication operation ⊗, and the division operation ⊘ on à and B˜ are expressed as[20,23]:

A˜B˜=(al+bl,ac+bc,ar+br)
A˜B˜=(albr,acbc,arbl)
A˜B˜=(albl,acbc,arbr)
A˜k=(kal,kac,kar)
A˜B˜=(al/br,ac/bc,ar/bl)

A crisp number k can be regarded as (k, k, k). Therefore, based on the GMIR, the addition operation and the subtraction operation between a triangular fuzzy number à = (al, ac, ar) and a crisp number b are defined as:

P(A˜k)=P[(al+k,ac+k,ar+k)]=al+4ac+ar6+k
P(A˜k)=P[(alk,ack,ark)]=al+4ac+ar6k

2.1.5 Ranking of triangular fuzzy numbers and crisp numbers by GMIR

The problem of ordering fuzzy numbers has been addressed by many researchers and consequently many fuzzy ranking methods exist. In this paper, the GMIR method is used to define the ranking of triangular fuzzy numbers and crisp numbers.

Let à = (al, ac, ar) be a triangular fuzzy number and k be a crisp number. Based on the GMIR method, the ranking number has the following properties:

  • P(Ã) > k if and only if Ãk

  • P(Ã) ≥ k if and only if Ãk

  • P (Ã) = k if and only if Ãk

  • P (Ã) < k if and only if Ãk

  • P (Ã) ≤ k if and only if Ãk

The notations Ãk, Ãk, Ãk, Ãk, Ãk signify that à has a higher ranking than k, at least the same ranking as k, the same ranking as k, a lower ranking than k and at most the same ranking as k respectively. It is also easy to determine the ranking orders among multiple triangular fuzzy numbers in a similar way.

2.2. Representation of the customer satisfaction value

For the distribution of durable goods such as books and consumer electronic devices, many VRP models aim to minimize the distribution costs and use the time window to represent the customers’ time constraints. However, short product lives and fast transportation are two characteristics of the supply chain for fresh and perishable foods[4]. Chen suggested using an S-shaped curve to depict the relationship between the freshness of perishable food products and profit[24]. Therefore, a Z-shaped curve is adopted here to depict the relationship between customer satisfaction and arrival time as was suggested in Ref. 12.

Fig. 2 gives a graphical method for illustrating this curve. For example, let ai = 2 and bi = 6. When the arrival time of the vehicle to customer i is within two hours, the customer will be very satisfied. As the arrival time passes two hours, the customer satisfaction value (csv) starts to decrease. After six hours, the customer will be very dissatisfied with the delivery.

Fig. 2.

Z-Shaped customer satisfaction function.

To describe the Z-shaped curve mathematically, Eq. (1) is used to depict the customer satisfaction value based on the built-in membership function in Matlab[25].

CSV(t)={1tai12(taibiai)2aitai+bi22(tbibiai)2ai+bi2tbi0tbi

2.3. Notations and model formulation

The proposed model consists of the following indices, parameters, and decision variables:

  • -

    An index k for the set of m vehicles V = {v1, v2, …, vm}.

  • -

    Indices h, i, j for a set of n nodes with customers C = {c1, c2, …, cn}.

  • -

    C is divided into u mutually exclusive nonempty subsets C = {C1,C2, …, Cu}, each subset representing a cluster of customers.

  • -

    C+ = C ∪ {c0} (where c0 represents the distribution centre).

  • -

    Perishable food products are broadly categorized into s types according to their temperature demands for storage. Correspondingly, there should be s types of cabins with different temperature values. The index z is used for the set of temperature values E = {e1, e2, …, es}.

  • -

    Customer demand for z types of perishable food products for customer i is diz.

  • -

    The load capacity of each cabin is q.

  • -

    The vehicle capacity is denoted by the set B. For example, vehicle k can accommodate Bk cabins.

  • -

    pkz cabins are used to store the z type of perishable food products in vehicle k.

  • -

    The satisfaction value of the customer i is csvi, which is generated using Eq. (15). The left extreme and right extreme of the i-th customer satisfaction curve are ai and bi respectively.

  • -

    Given that the travel time between customer i and customer j cannot be predicted precisely[12,26], it is considered as a triangular fuzzy number t˜ij=(tijl,tijc,tijr).

  • -

    The moment when customer i begins to be served by any vehicle is denoted by the time variable ti (where t0 represents the time when vehicles depart from the distribution centre).

  • -

    Service time at node i is denoted by wi.

  • -

    M is a large positive number.

  • -

    If there exists a vehicle that travels from customer i to customer j (ij), then xijk=1; otherwise xijk=0.

  • -

    If customer i is served by vehicle k, then yik=1; otherwise yik=0.

The mathematical formulation for the clustered MTJD can be stated as follows:

maxZ=i=1nCSV(ti)
s.t.
k=1mi=1nxijk=1,jC
k=1mi=1nxjik=1,jC
i=1nx0ik=1,kV
i=1nxi0k=1,kV
k=1myik=1,iC
yikyikxijk,i,jC,kVk
i=1nx0ikyik=1,kV
i=1nxi0kyik=1,kV
i=0nxihkj=0nxhjk=0,hC,kV
t0i˜˜tj,iC,jC
ti+witij˜M(1xijk)tj,iC+,jC,kV,ij
z=1spkzs,kV
i=1ndizyikpkzq,kV,zE
z=1spkzBk,kV
CiCj=i,j,ij
xijk{0,1},i,jC+,kV
yik{0,1},iC,kV
ti0iC+

The objective function (16) aims to maximize the sum of the customer satisfaction value. Eq. (17) and Eq. (18) ensure that all customers are served only once. Eq. (19) and Eq. (20) indicate that each route starts and ends at the distribution centre and only one trip is allowed. Eq. (21) states that each customer is served by only one vehicle. Eqs. (22)(24) are used to represent the relationship between xijk and yik. Eq. (25) guarantees that flow conservation is satisfied. Eq. (26) and Eq. (27) guarantee the consistency of time variables. Eq. (28) indicates that each vehicle accommodates all types of cabins. Eq. (29) guarantees that the total capacity of cabins in any vehicle for accommodating the z type of perishable food products is not exceeded. Eq. (30) ensures that the aggregated capacity of refrigerated containers does not exceed the loading capacity of the vehicle. Eq. (31) guarantees that each customer node belongs to only one cluster. Eqs. (32)(34) give range constraints on the decision variables.

3. Solution Approaches

3.1. Coding method

Each firefly represents a possible solution. For any firefly, its position can be denoted as the set {(vk, ci, ti)|iC, kV}. This indicates that the elapsed time when vehicle vk arrives at customer node ci is ti. The light intensity of firefly i is denoted by Ii, which is the total customer satisfaction value.

3.2. Fuzzification and defuzzification

Based on Eq. (7), the GMIR of the fuzzy travel time tij˜=(tijl,tijc,tijr) is defined via Eq. (35).

P(tij˜)=16[tijl+4tijc+tijr]

After the algorithm reads the instance data, tijc is generated via the distance between node i and node j divided by the speed. Next, tijl and tijr are generated based on the value of tijc. The relationship among the three numbers can be determined by the decision maker. Finally, the triangular fuzzy number tij˜=(tijl,tijc,tijr) is defuzzified by applying Eq. (35).

3.3. Population initialization

In many meta-heuristic algorithms including some variants of the firefly algorithm, the initial population is generated randomly[16,27,28]. Thus, pop_size is defined as the population size, and all fireflies F={F0, F1, …, Fpop_size-1} are generated at random. Although the initial population of reasonably structured solutions lacks sufficient diversity, there is also a possibility that it could generate high-quality near-optimal values[29]. For example, Refs. 15 and 17 choose items with high value density to generate initial solutions. Therefore, Algorithm 1 is designed to generate the initial firefly population with structured solutions. Additionally, experiments are carried out with the initial fireflies generated both randomly and by using Algorithm 1 for the purpose of comparison.

1:
for i =0: pop_size-1
2:
  Allocate m neighbouring nodes of the distribution centre from different clusters to each vehicle respectively;
3:
  while not all of the customer nodes are visited
4:
    for j=0: m−1
5:
      if r=rand[0,1] ≤ PI&&all constraints are satisfied
6:
        Allocate the nearest unvisited customer node to vehicle j;
7:
      end if
8:
    end for
9:
  end while
10:
Initialize light intensity Ii;
11:
end for i
12: Find the current brightest firefly Fopt;

Algorithm 1

To better present Algorithm 1, two examples are shown in Fig. 3. As shown in Fig. 3(a), for node 1, if the random number r is less than or equal to PI, an edge is generated from itself to its nearest neighbor – node 4. Fig. 3(b) shows the opposite situation. For node 1, if the random number r is larger than PI, it will give up connecting to its nearest neighbour and wait for the next turn. Thus, node 4 will be connected with node 2 and therefore a solution which differs from that in Fig. 3(a) will be generated. From the above-mentioned presentation, it can easily be inferred that the uniform random variable PI can prevent the generation of multiple duplicate solutions.

Fig. 3.

Examples of population initialization.

3.4. Movement of fireflies

Given that discrete optimization problems cannot be solved by the original firefly algorithm which was designed for coping with continuous optimization problems, several versions of the discrete firefly algorithm have been proposed[1517]. In this algorithm, the arrival time series are used to measure the distance between two fireflies. For example, let ti=(t1i,,tni) be the arrival time series vector of firefly i. In this sequence, the arrival time of a vehicle to customer h is denoted by thi. Thus, the distance between any two fireflies Fi and Fj is determined by

rij=|h=1n(thjthi)|

In the basic firefly algorithm, γ denotes the light absorption coefficient. It is crucial to the convergence speed of the algorithm[13]. Based on Eq. (36), the movement of firefly Fj toward firefly Fi is determined by

n=rij/γ
Fj=InsertionFunction(Fj,n)

Eq. (37) indicates that the length of the movement of a firefly will be the smallest integral value that is greater than or equal to rij/γ. The InsertionFunction is similar to that given in Ref. 16. It selects and extracts one random node from a random route. After that, this node is randomly moved beside a node for which the cluster is same as the extracted one. Fig. 4 shows two examples of how the InsertionFunction works. In Fig. 4(a), node 3 is randomly selected to be moved. Node 11, which belongs to same cluster, is randomly selected as the target node. Next, node 3 is moved to the left of node 11 if the random variable p is less than or equal to 0.5; otherwise it is moved to the right of node 11. As shown in Fig. 4(b), if the extracted node and the target node belong to the same route, the method is also viable.

Fig. 4.

Examples of the InsertionFunction.

Instead of using a repair operator, capacity constraint is taken into account to avoid generating infeasible solutions. This function is repeated n times to enable a movement of firefly Fj. If the new position is superior to the original position, then the firefly is updated.

If Fj is the firefly with the largest light intensity, it moves randomly[30] in the basic firefly algorithm. Tilahun and Ong pointed out that this behaviour may lead to a reduction in the algorithm performance, because the light intensity of the brightest firefly may decrease in certain directions. They suggested randomly choosing a direction in which the brightness of the brightest firefly increases if the firefly moves in that direction. If there exists no such direction, the brightest firefly must stay in its current position[31]. However, the brightest firefly Fj will cease moving in the discrete firefly algorithm proposed here, because rjj = 0 and consequently n = 0 by applying Eq. (36) and Eq. (37). Therefore, n = 1 is set in Eq. (38) for the brightest firefly.

3.5. Repair operation

Sometimes, several customer nodes may exist which are at the end of the route with a csv of zero. The approach mentioned above may generate an impractical route for these, because the visit order does not affect the objective value of the solution. Considering the two routes in Fig. 5 for example, it can be seen that the distance of route 1 is clearly longer than that of route 2, but the sum of csv is the same for the two routes. Therefore, route 1 should be modified by reordering the visiting sequence.

Fig. 5.

An example of the repair operation.

Assume the csv of the successor of node h is zero and the index of the route to be repaired is k. To remedy the problem mentioned above, Algorithm 2 is proposed and presented as follows.

1:
Set J={j|CSV(tj)=0,yjk=1};
2:
Set p = h, yjk=0, xijk=0, tj = 0, ∀iC, ∀jJ;
3:
repeat
4:
  Randomly select an element s from {j|tpjtpl, ∀lJ, ∃ jJ};
5:
  Connect node p and node s;
6:
  Calculate ts; Set xpsk=1; Set ysk=1; Set p = s; Set J = J −{s};
7:
until J = 

Algorithm 2

3.6. Procedure for the discrete firefly algorithm

As in the basic firefly algorithm, the movement procedure is repeated until the best solution remains unchanged over rep_gen iterations. To sum up, the pseudocode of the proposed discrete firefly algorithm (DFA) can be summarized according to Algorithm 3.

1: Read instance data (customer locations, customer demands, vehicle capacity, travel times, etc.);
2:
Set the parameters of the discrete firefly algorithm;
3:
for i=0: n
4:
  for j=0: n
5:
    if ij
6:
      Fuzzify tijc and defuzzify t˜ij using Eq. (35);
7:
    else
8:
      tij=0;
9:
    end if
10:
  end for
11:
end for
12:
Initialize the firefly populations;
13:
repeat
14:
  for i=0: pop_size-1
15:
    for j =0: pop_size-1
16:
      if ij&& Ii >Ij
17:
        Move Fj using Eqs. 36–38;
18:
      end if
19:
    end for
20:
  end for
21:
  Find Fopt;
22:
  Move Fopt using Eq. (38) (n=1
);
23:
until Fopt remains unchanged over rep_gen iterations
24:
for k=0: m-1
25:
  if a node with zero csv is found in route k
26:
    Repair Fopt using Algorithm 2;
27:
  end if
28:
end for

Algorithm 3

4. Simulation Experiment

In this section, experiments are carried out to validate the proposed mathematical formulation and algorithm. All computational experiments are coded with Visual C# in a PC with a 3.2GHz Intel Pentium G3420 processor and 8GB RAM.

4.1. Data generation

In the experiments, problems are generated to evaluate the performance of the proposed algorithm. In each instance, the values s=5, q=160L, γ =50, t0=0, rep_gen=2000 and speed=500m/minute are set. All the customer nodes have corresponding (x,y) coordinates, where both x and y are between −25000m and 25000m. Parameters wi, ai, and bi which refer to time are measured in minutes. In the numerical example about joint distribution in Ref. 10, perishable food products are measured in litres and are categorized into five groups according to their temperature ranges[10]: (1) tuna, (2) ice cream, ice cubes, and frozen dumplings, (3) milk and juice; (4) biscuits and medicines and (5) lunch boxes and hot meals. Therefore, s is set at a value of 5 and diz is measured in litres in this example. According to general consumer behaviour, customer demand for each type of perishable food product is generated with a uniform distribution, where di1~U(0,5), di2~U(0,8), di3~U(0,10), di4~U(0,15) and di5~U(0,10). As a matter of experience, it may be satisfactory for many consumers to receive the food products within about two-and-a-half hours, and a delivery time of over about six hours could result in customer dissatisfaction. Given that customer attitudes differ with regard to delivery time, the variances of the customer satisfaction parameters are at 900, which can take into account the diversity of customer attitudes appropriately. Therefore, the parameters ai and bi are generated with normal distributions N(150,900) and N(360,900) respectively. Given that five minutes or so is enough time for serving a consumer, the parameter wi is generated with normal distribution N(5,1). Travel times are fuzzified as triangular fuzzy numbers with tijl=0.9tij and tijr=1.1tij. The starting point for each vehicle is (0,0). Given that 10–25 fireflies can deal with most applications[32], pop_size was set to 20.

4.2. Experimental results

The proposed model and algorithm was applied to 22 instances. Each of these instances was run 10 times. The experimental results are shown in Table 1. DFA1 was implemented with a randomly generated initial population, and DFA2 was implemented with an initial population generated by Algorithm 1. Since that the mathematical model of the clustered MTJD is first proposed in this paper, there are no pre-existing dedicated algorithms. Given that Algorithm 1 is a type of greedy algorithm which connects the current node with the nearest unvisited node at each stage when PI is set to 100%, it can be used as the benchmark for evaluating the effectiveness of the solution approaches, because a greedy algorithm is simple to develop and can solve many applications including the travelling salesman problem and the vehicle scheduling problem[33].

No. n u m B DFA1 DFA2 Greedy Algorithm

best median worst Std.dev best median worst Std.dev
1 30 4 1 (6) 14.320 12.877 12.067 0.903 13.130 12.960 12.097 0.465 11.795
2 30 4 2 (5,5) 27.245 26.553 24.757 1.020 27.389 27.245 26.385 0.302 25.021
3 30 6 1 (8) 14.680 13.156 11.330 1.335 14.059 14.055 14.050 0.005 14.037
4 30 6 2 (5,5) 26.921 25.641 24.564 0.901 26.908 26.630 26.524 0.157 20.083
5 60 4 1 (15) 23.509 22.237 21.427 0.680 19.216 19.092 18.968 0.131 18.765
6 60 4 2 (8,7) 40.476 36.761 33.766 1.792 39.605 38.583 36.461 1.270 34.015
7 60 4 3 (5,5,6) 53.864 52.146 48.698 1.879 51.402 50.530 49.011 0.721 45.517
8 60 6 1 (15) 21.217 20.458 18.861 0.644 20.869 20.869 20.568 0.101 20.557
9 60 6 2 (8,7) 42.481 37.911 34.768 2.145 42.014 38.515 37.865 1.241 37.428
10 60 6 3 (5,5,6) 53.565 51.927 48.908 1.662 53.293 52.196 51.212 0.605 47.083
11 90 4 2 (11,10) 49.032 47.215 46.343 0.796 45.525 44.247 43.846 0.601 40.681
12 90 4 3 (7,7,8) 68.326 66.570 61.686 1.928 63.795 61.877 58.119 1.809 52.656
13 90 4 4 (6,6,6,6) 84.826 82.958 77.088 2.371 80.670 78.474 78.040 0.985 70.656
14 90 6 2 (11,10) 42.913 40.997 38.602 1.443 41.073 39.747 39.737 0.420 39.202
15 90 6 3 (7,7,8) 58.706 58.011 57.384 0.452 58.303 57.374 54.761 1.165 52.687
16 90 6 4 (6,6,6,6) 78.424 72.977 69.619 2.981 77.733 76.826 74.740 1.088 69.922
17 120 4 3 (8,8,9) 71.352 67.656 60.933 2.935 69.584 68.273 64.361 2.019 58.110
18 120 4 4 (7,7,7,7) 97.067 92.987 88.357 2.245 88.341 85.315 82.112 1.244 71.950
19 120 4 5 (5,6,6,6,6) 109.326 104.839 97.800 3.205 110.084 107.392 105.322 1.611 85.931
20 120 6 3 (8,8,9) 66.527 62.470 58.152 2.270 64.457 63.261 62.307 0.874 60.261
21 120 6 4 (7,7,7,7) 84.652 80.586 76.911 1.461 88.977 87.367 85.957 0.599 82.972
22 120 6 5 (5,6,6,6,6) 105.389 102.865 92.887 3.660 105.982 103.702 100.049 1.464 93.592
Table 1.

Computational results and comparisons for the three algorithms in 22 problem instances.

According to the results presented in Table 1, DFA1 outperforms DFA2 in terms of the best value for all the instances except for instances 2, 19, 21 and 22. The median values obtained by DFA2 are better than those generated by DFA1 for instances with 120 customer nodes, except for instance 18. All the solutions obtained by DFA2 are better than those obtained by the greedy algorithm, but the worst value of 10 instances and the mean value of three instances obtained by DFA1 are even worse than those obtained by the greedy algorithm. Therefore, it can be inferred that DFA1 is suitable for solving small and medium-sized problem instances. It is also worth mentioning that DFA2 presents stronger robustness than DFA1 in general, as shown by the standard deviation values. Additionally, for customers with identical numbers and clusters, more vehicles contribute to higher customer satisfaction values, which partially proves the effectiveness of the algorithms developed in this paper.

Fig. 6 (some dots are overlapping) and Fig. 7 present detailed results of several instances, in order to further investigate the effectiveness of the proposed algorithm. Fig. 8 and Table 2 reveal the efficiency of the algorithm over instances 6 and 16. It can be clearly observed that a larger population size consumes more computational time for both versions of the discrete firefly algorithm proposed in this paper.

Fig. 6.

Graphical description for best operation routes over four instances with pop_size=20.

Fig. 7.

Best vehicle load planning results over No.6 instance.

Fig. 8.

Convergence curves of the discrete firefly algorithms over instance 6.

Algorithm No. pop_size Zbest Zmedian tmeidan(s)
DFA1 6 10 40.906 37.128 105
DFA1 6 20 40.476 36.761 336
DFA1 6 30 41.617 38.837 746
DFA1 6 40 41.622 38.152 1089
DFA1 16 10 77.842 73.152 167
DFA1 16 20 78.424 72.977 406
DFA1 16 30 78.179 72.889 709
DFA1 16 40 78.896 74.685 1478
DFA2 6 10 39.605 39.023 68
DFA2 6 20 39.605 38.583 203
DFA2 6 30 41.564 39.384 444
DFA2 6 40 41.082 37.701 656
DFA2 16 10 78.025 76.288 138
DFA2 16 20 77.733 76.826 595
DFA2 16 30 77.877 76.261 785
DFA2 16 40 78.257 77.261 1739
Table 2.

Small study of solution quality and average runtime of two instances with different population size.

5. Conclusions

In this paper, a mathematical model for a clustered multi-temperature joint distribution with fuzzy travel times was proposed. It can be applied to solve the distribution of perishable food products with different temperature demands. In contrast to existing MTJD models, the proposed model aims at maximizing the customer satisfaction value by using a Z-shaped function, regards travel times as triangular fuzzy numbers, and considers a heterogeneous set of vehicles with different capacities.

Due to the complexity of the problem, the paper also considers the design of two versions of the discrete firefly algorithm which differ in the population initialization method but share the same movement scheme for fireflies. To validate the two algorithms, 22 problem instances are considered. The simulation indicates that the discrete firefly algorithms developed can effectively solve the clustered MTJD.

Future research may consider integrating the clustered MTJD with inventory optimization, time-varying networks, or the hub location problem. Sensitivity analysis can be conducted by comparing fuzzy parameters and optimal solutions in alternative defuzzification techniques such as graded k-preference integration and α-cuts. Establishing multiple objectives to increase the practicability of this model is also a possible direction for future research.

Acknowledgement

This research was funded by Beijing Jiaotong University scientific research projects T14L00360 and T14L00070.

References

9.YJ Cho and CC Li, Application of Multi-temperature Refrigerated Container to Improve the Distribution of Cold Logistics, J. East. Asia Soc. Transp. Stud, Vol. 6, 2005, pp. 2794-2808.
12.S Lu and X Wang, From a Literature Review to a Theoretical Framework for Cloud-Based Internet of Things and Optimization Enabled Perishable Food Cold Chain Management, Advances in Energy Science and Equipment Engineering II, CRC, Leiden, 2017. In Press.
22.SH Chen, ST Wang, and SM Chang, Some Properties of Graded Mean Integration Representation of L-R Type Fuzzy Numbers, Tamsui Oxford J. Math. Sci, Vol. 22, No. 2, 2006, pp. 185-208.
24.H Chen and L Fang, Research on shelf life-constrained production scheduling for perishable food, J. Guangxi U.: Nat. Sci. Ed, Vol. 38, No. 3, 2013, pp. 729-737.
Journal
International Journal of Computational Intelligence Systems
Volume-Issue
11 - 1
Pages
195 - 205
Publication Date
2018/01/01
ISSN (Online)
1875-6883
ISSN (Print)
1875-6891
DOI
10.2991/ijcis.11.1.15How to use a DOI?
Copyright
© 2018, the Authors. Published by Atlantis Press.
Open Access
This is an open access article under the CC BY-NC license (http://creativecommons.org/licences/by-nc/4.0/).

Cite this article

TY  - JOUR
AU  - Sichao Lu
AU  - Xifu Wang
PY  - 2018
DA  - 2018/01/01
TI  - Discrete Firefly Algorithm for Clustered Multi-Temperature Joint Distribution with Fuzzy Travel Times
JO  - International Journal of Computational Intelligence Systems
SP  - 195
EP  - 205
VL  - 11
IS  - 1
SN  - 1875-6883
UR  - https://doi.org/10.2991/ijcis.11.1.15
DO  - 10.2991/ijcis.11.1.15
ID  - Lu2018
ER  -