Advertisement

Novel Computational Intelligence for Optimizing Cyber Physical Pre-evaluation System

Chapter
Part of the Studies in Computational Intelligence book series (SCI, volume 540)

Abstract

Owing to the quality heterogeneity of returned used products, firms engaged in re-manufacturing activities are obliged to employ 100 % inspection of these products to evaluate their quality and suitability for re-manufacturing. In addition to visual inspection, a recent tendency is to use data recorded in electronic devices (e.g., radio frequency identification (RFID)) implanted in the products. In this way, information is obtained quickly without the need for complete (and expensive) product disassembly. Nevertheless, making sense of RFID data in a complex cyber physical system (CPS) environment (which involves such as cloud computing for used product life cycle information retrieval and physically used products scanning) is a complex task. For instance, if an RFID readers fails, there may be missing values exist. The purpose of this chapter is to employ two computational intelligence (CI) optimization methods which can improve the reliability of such inspection process.

Keywords

Re-manufacturability Cyber physical pre-evaluation system Reliability-redundancy allocation problem Firefly algorithm Teaching–learning-based optimization Radio frequency identification 

1 Introduction

Returns acquired by a re-manufacturer are typically highly variable in quality [16]. A significant consequence of this uncertainty is the inclusion of a collection/classification stage and a corresponding system of quality-dependent routing of supply in a reverse logistics network [13]. The potential value of sorting and classification product returns has been explored by different researchers such as [7] and several sorting policies have also been proposed in the literature (e.g., [4]). In addition, the management of product return is characterized also by the lack of information associated with such used products [29]. The recent emergence of networked radio frequency identification (RFID) system is a means of connecting a product tagged with an RFID chip to a network and thereby carrying complete information associated with it throughout its lifecycle. In this way information is obtained quickly without the need for complete (and expensive) product disassembly [53]. Several authors (e.g., [14, 27, 34]) have mentioned the use or potential use of RFID and related technology in the reverse logistics network. Nevertheless, making sense of RFID data is a complex task. For instance, if an RFID readers fails, there may be missing values exist [10]. The purpose of this chapter is to employ two innovative computational intelligence (CI) approaches for improving the reliability of such classification/inspection process.

The remainder of this chapter is organized as follows. Subsequent to the introduction in Sect. 1, the background of cyber physical Re-manufacturability pre-evaluation system is briefed in Sect. 2. Then, the problem statement is presented in Sect. 3 which is followed by a problem formulation detailed in Sect. 4. The proposed methodologies are then detailed in Sect. 5. Next Sect. 6 conducts an experimental study to demonstrate the feasibility of our proposed approaches. The future research directions are highlighted in Sect. 6. Finally, the conclusion is drawn in Sect. 6 of this chapter.

2 Background of Cyber Physical Pre-evaluation System

2.1 What is Re-manufacturability?

There is a growing interest in re-manufacturability analyses of the re-manufacturing systems since it is the key element to maintain customer satisfaction and thus company profitability. Generally speaking, re-manufacturability is the ability of used products to be easily re-manufactured and be determined by the configurable parameters, the failed state, and the re-manufacturing technology [49]. Regarding the configurable parameters, Wu [42] pointed out that the influent factors included the technological feasibility of re-manufacturing, the economic feasibility of re-manufacturing, the environmental feasibility of re-manufacturing, and the product’s service ability. In a similar vein, [12] emphasized that for evaluating the re-manufacturability of used products, an integrated method in which the technology feasibility (including disassembly, cleaning, inspection and sorting, par reconditioning, machine upgrading and reassembly), economic feasibility (focusing on the re-manufacturing cost), and environmental benefits (such as energy saving, material saving and pollution reduction) should be analysed. Furthermore, some researchers (e.g., [2, 8, 51]) proposed that to enhance re-manufacturability of used products, manufacturers should take into account the early stages of the products’ designs. In the light of this statement, in [17, 38, 40], the authors stated that different design structure matrix could be used as a very useful tool to examine the relationship between the different processes in order to obtain a clear ranking of the easily activities of re-manufacturing.

2.2 Why Re-manufacturability Pre-evaluation?

In most cases, re-manufacturing processes must adopt the activity of pre-evaluation because products have not been designed to be re-manufactured [52]. This activity has extracted the “secret” affecting the success of re-manufacturing since it allows for the selective using of desired parts and/or materials. In other words, it provide a relatively efficient and effective means for a re-manufacturer to obtain feedback before the used products are admitted into the re-manufacturing plant, specially, the information about which used products/parts can be disassembled [43].

2.3 Cyber Physical Pre-evaluation System

One of the ways to evaluation such ability is through cyber-physical system (CPS), which, in our context, use sensor-embedded products with networked computing to control the evaluation processes in order to remove uncertainty to the re-manufacturing systems. An early of the successful marriage of sensor-based products and evaluation processes is radio frequency identification (RFID) tag. The advent of RFID tag is critical to automatic identification, movement tracking, access control, information collection, and evaluation of operation/system’s performance. Furthermore, it is also considered by some researchers (e.g., [23, 24, 25, 29]) as one of the most technology for revolutionizing a wide range of applications including re-manufacturing and reverse logistics.

3 Problem Statement

When a used product is collected, the first step is to evaluate its re-manufacturability, which is the premise to decide whether it is worthy to re-manufacture the product. At this stage, effective and reliable systems are required to gather and evaluate product usage data. Recently, some studies (e.g., [1, 28, 53]) have reported permanent sensor embedded tagging (such as RFID) may generate valuable information for improving the efficiency of re-manufacturing process. Their analyses suggested that, since there is a high level of uncertainty about the quality of components entering the re-manufacturing process, RFID-derived information can assist in sorting components where manual inspection is traditionally employed.

This may be true for parts that are sensitive to re-manufacture, however, one of the major puzzles that RFID classification system has posed for practitioners is the reading accuracy and system reliability after its adoption. Bearing this in mind, in this study, we are about to set up a 4-stage inspection procedure to keep the misclassification rate [53] at the lowest level. Each stage is constituted of an RFID inspecting system that is responsible for a certain type of data collection and evaluation. While the used products flow passing through these four inspection points, if any of them works improperly, the operator of the cross docking should be notified to take that certain used product out for a further inspection. Such interruption highly affects the working efficiency of a cross docking station within a re-manufacturing process and thus the reliability of the entire RFID system should be enhanced as much as possible by taking various constraints into account.

4 Mathematical Modelling

As it can be seen, our focal problem falls under the category of RFID operational level research. Nevertheless, according to our recent review [44], literature provides little guidance in addressing this issue. Therefore, in this chapter, we propose to model our focal problem as a reliability-redundancy allocation problem (RRAP).

4.1 Reliability-Redundancy Allocation Problem

The optimization of system reliability is very important in many real applications and has attracted increasing attention in academic field and a variety of engineering fields. Typically RRAP problems involve the selection of components with multiple choices and redundancy levels that produce maximum benefits, and are subject to many constraints such as the cost, weight, and volume. In general an RRAP problem can be formulated as a mixed-integer programming problem which is shown in Table 1 [41].
Table 1

General form of RRAP problem

General Form of RRAP Problem

Maximize

 
 

R s  = f(rn)

Subject to

 
 

\( \begin{array}{*{20}c} {g_{j} \left( {{\mathbf{r}},{\mathbf{n}}} \right) \le l_{j} ,} & {j = 1,2, \cdots ,m} \\ \end{array} \)

where:

 

R s

the reliability of a system

r = (r 1r 2, ···, r d )

the vector of the component reliabilities for the system

r i :

the reliability of the ith subsystem

n = (n 1n 2, ···, n d )

the vector of redundancy allocation for the system

n i

the number of components in the ith subsystem

f( · )

the objective function for the overall system reliability

g(rn)

the set of constraint functions

g j (rn)

the jth constraint function

d

the number of subsystems in the system

l

the vector of resource limitations

The goal of RRAP problem is to find the optimal combination of components and the reliabilities of the components to achieve the highest system reliability [41].

4.2 4-Stage Series System

Having the characteristics of our targeted RFID system in mind (see Sect. 2 for more details), we decide to formulate it as a 4-stage series system shown in Eq. (1). In the literature, this formulation has also been successfully used for modelling over-speed protection system of a gas turbine [21, 50].
$$ \text{Maximize}:\,f\left( {{\mathbf{r}},{\mathbf{n}}} \right) = \prod\limits_{i = 1}^{4} {\left[ {1 - \left( {1 - r_{i} } \right)^{{n_{i} }} } \right]} $$
(1)
Subject to:
$$ g_{1} \left( {{\mathbf{r}},{\mathbf{n}}} \right) = \sum\limits_{i = 1}^{4} {v_{i} n_{i} } \le V, $$
$$ g_{2} \left( {{\mathbf{r}},{\mathbf{n}}} \right) = \sum\limits_{i = 1}^{4} {\alpha_{i} \left( {{{ - T} \mathord{\left/ {\vphantom {{ - T} {\ln \left( {r_{i} } \right)}}} \right. \kern-0pt} {\ln \left( {r_{i} } \right)}}} \right)^{{\beta_{i} }} } \left[ {n_{i} + e^{{0.25n_{i} }} } \right] \le C, $$
$$ g_{3} \left( {{\mathbf{r}},{\mathbf{n}}} \right) = \sum\limits_{i = 1}^{4} {w_{i} n_{i} e^{{0.25n_{i} }} } \le W. $$
where 0.5 ≤ r i  ≤ 1 − 10−6, r i  ∊ R +, 1 ≤ n i  ≤ 10, n i  ∊ Z +, v i represents the volume of each component in the subsystem i, V denotes the upper limit on the sum of the subsystems’ products of volume, C is the upper limit on the cost of the system, \( \alpha_{i} \left( {{{ - T} \mathord{\left/ {\vphantom {{ - T} {\ln \left( {r_{i} } \right)}}} \right. \kern-0pt} {\ln \left( {r_{i} } \right)}}} \right)^{{\beta_{i} }} \) is the cost of each component with reliability r i at subsystem i in which α i and β i are coefficients, T is the operating time during which the component must not fail, and W is the upper limit on the weight of the system.

5 Proposed Methodology

Many classical mathematical methods have failed to address the non-convexities and non-smoothness in RRAP problems. As an alternative to the classical optimization approaches, the CI approaches have been given much attention by many researchers because of their superior capability in finding an almost global optimal solution. In this research, we choose teaching—learning-based optimization (TLBO) and firefly algorithm (FA) as a vehicle to address our 4-stage series system problem.

5.1 Background of TLBO

Teaching—earning-based optimization (TLBO) is a new efficient population based algorithm inspired by the influence of a teacher on the output of learners in a class, which learners first acquire knowledge from a teacher (i.e., teacher phase) and then from classmates (i.e., learner phase) [35]. In principle, population consists of learners in a class and design variables are courses offered. The output in TLBO algorithm is considered in terms of results or grades of the learners which depend on the quality of teacher. That means, a high quality teacher is usually considered as a highly learned person who trains learners so that they can have better results in terms of their marks or grades. Moreover, learners also learn from the interaction among themselves which also helps in improving their results. Working of both the phase is explained below.

Teacher Phase: In the model, this phase produces a random ordered state of points called learners within the search space. Then a point is considered as the teacher, who is highly learned person and shares his or her knowledge with the learners, and others learn significant group information from the teacher. It is the first part of the algorithm where the mean of a class increases from M A to M B depending upon a good teacher. At this point, we assumed a good teacher is one who brings his/her learners up to his/her level in terms of knowledge. However, in practice this is not possible and a teacher can only move the mean of a class up to some extent depending on the capability of the class. This follows a random process depending on many factors [35]. Let M i be the mean and T i be the teacher at any iteration i. T i will try to more mean M i towards its own level, so now the new mean will be T i designated as M new . The solution is updated according to the difference between the existing and the new mean given by Eq. (2) [35]:
$$ Difference\_Mean_{i} = r_{i} \left( {M_{new} - T_{F} M_{i} } \right) $$
(2)
The value of T F can be either 1 or 2, which is again a heuristic step and decided randomly with equal probability as shown in Eq. (3) [35]:
$$ T_{F} = round\left[ {1 + rand\left( {0,1} \right)\left\{ {2 - 1} \right\}} \right]. $$
(3)
This difference modifies the existing solution according to Eq. (4) [35]:
$$ X_{new,i} = X_{old,i} + Difference\_Mean_{i} $$
(4)
Learner Phase: It is the second part of the algorithm where learners increase their knowledge by interaction among themselves. So, a solution is randomly interacted to learn something new with other solutions in the population. In the light of this statement, a solution will learn new information if the other solutions have more knowledge than him or her. Mathematically the learning phenomenon of this phase is expressed in Eq. (5) [35]:
$$ \begin{gathered} X_{new,i} = X_{old,i} + r_{i} \left( {X_{i} - X_{j} } \right), \, if \, f\left( {X_{i} } \right) < f\left( {X_{j} } \right) \hfill \\ X_{new,i} = X_{old,i} + r_{i} \left( {X_{j} - X_{i} } \right), \, if \, f\left( {X_{j} } \right) < f\left( {X_{i} } \right) \hfill \\ \end{gathered} $$
(5)
After a number of sequential teaching–learning cycles, where the teacher convey knowledge among the learners and those level increases toward his or her own level, the distribution of the randomness within the search space becomes smaller and smaller about to point considering as teacher. It means knowledge level of the whole class shows smoothness and the algorithm converges to a solution. Also, a termination criterion can be a predetermined maximum iteration number is reached.

In many aspects, TLBO resembles evolutionary algorithms [33] such as an initial population is randomly generated; moving/learning towards teacher and classmates can be regarded as a special mutation operator; and selection is deterministic (i.e., two solutions are compared and the better one always survives) [11 ]. The TLBO algorithm has been used in solving many problems, remarkable results have been reported about TLBO outperforming many algorithms such as differential evolution [39], evolutionary strategies [6], and particle swarm optimization [26].

5.2 Background of FA

The firefly algorithm (FA) is a meta-heuristic, nature-inspired, optimization algorithm which is based on the social (flashing) behaviour of fireflies, or lighting bugs, in the summer sky in the tropical temperature regions [45, 46, 47]. It was developed by Dr. Xin-She Yang at Cambridge University in 2007, and it is based on the swarm behaviour such as fish, insects, or bird schooling in nature. In the FA, physical entities (fireflies) are randomly distributed in the search space. They carry a bio-luminescence quality, called luciferin, as a signal to communicate with other fireflies, especially to prey attractions [5]. In detail, each firefly is attracted by the brighter glow of other neighbouring fireflies. The attractiveness decreases as their distance increases. If there is no brighter one than a particular firefly, it will move randomly. Its main advantage is the fact that it uses mainly real random numbers, and it is based on the global communication among the swarming particles (i.e., the fireflies), and as a result, it seems more effective in multi-objective optimization. Normally, FA uses the following three idealized rules [47] to simplify its search process to achieve an optimal solution:
  • Fireflies are unisex so that one firefly will be attracted to other fireflies regardless of their sex, that means no mutation operation will be done to alter the attractiveness fireflies have for each other;

  • The sharing of information or food between the fireflies is proportional to the attractiveness that increases with a decreasing Cartesian or Euclidean distance between them due to the fact that the air absorbs light. Thus for any two flashing fireflies, the less bright one will move towards the brighter one. If there is no brighter one than a particular firefly, it will move randomly;

  • The brightness of a firefly is determined by the landscape of the objective function. For the maximization problems, the light intensity is proportional to the value of the objective function.

Furthermore, there are two important issues in the FA that are the variation of light intensity or brightness and formulation of attractiveness. Yang [45] simplifies the attractiveness β of a firefly is determined by its brightness I which in turn is associated with the encoded objective function. As light intensity and thus attractiveness decreases as their distance from the source increases, the variations of light intensity and attractiveness should be monotonically decreasing functions.

  • Variation of Light Intensity: Suppose that there exists a swarm of n fireflies, and x i i = 1, 2, …, n represents a solution for a firefly i initially positioned randomly in the space, whereas f(x i ) denotes its fitness value. In the simplest form, the light intensity I(r) varies with the distance r monotonically and exponentially. That is determined through Eq. (6) [45, 46, 47]:
    $$ I = I_{0} e^{{ - \gamma r_{ij} }} $$
    (6)
    where I 0 is the original light intensity, γ is the light absorption coefficient, and r is the distance between firefly i and firefly j at x i and x j as Cartesian distance \( r_{ij} = \left\| {x_{i} - x_{j} } \right\| = \sqrt {\sum\limits_{k = 1}^{d} {\left( {x_{i,k} - x_{j,k} } \right)^{2} } } \)or the ℓ2 - norm, where x i,k is the kth component of the spatial coordinate x i of the ith firefly and d is the number of dimensions we have, for d = 2, we have \( r_{ij} = \sqrt {\left( {x_{i} - x_{j} } \right)^{2} + \left( {y_{i} - y_{j} } \right)^{2} } \).
  • Movement toward attractive Firefly: A firefly attractiveness is proportional to the light intensity seen by adjacent fireflies [45]. Each firefly has its distinctive attractiveness β which implies how strong it attracts other members of the swarm. However, the attractiveness is relative; it will vary with the distance between two fireflies. The attractiveness function β(r) of the firefly is determined via Eq. (7) [45, 46, 47]:
    $$ \beta = \beta_{0} e^{{ - \gamma r_{ij}^{2} }} $$
    (7)
    where, β 0 is the attractiveness at r = 0, and γ is the light absorption coefficient which controls the decrease of the light intensity.
  • The movement of a firefly i at location x i attracted to another more attractive (brighter) firefly j at location x j is determined based on Eq. (8) [45, 46, 47]:
    $$ x_{i} \left( {t + 1} \right) = x_{i} \left( t \right) + \beta_{0} e^{{ - \gamma r_{ij}^{2} }} \left( {x_{j} - x_{i} } \right) + \alpha \varepsilon_{i} $$
    (8)
    where, the first term is the current position of a firefly, the second term is used for considering a firefly’s attractiveness to light intensity seen by adjacent fireflies, and the third term is randomization with the vector of random variables ɛ i being drawn from a Gaussian distribution, in case there are not any brighter ones. The coefficient α is a randomization parameter determined by the problem of interest.
  • Special Cases: From Eq. (8), it is easy to see that there exit two limit cases when γ is small or large, respectively [45, 46, 47]. When γ tends to zero, the attractiveness and brightness are constant β = β 0 which means the light intensity does not decrease as the distance r between two fireflies increases. Therefore, a firefly can be seen by all other fireflies, a single local or global optimum can be easily reached. This limiting case corresponds to the standard particle swarm optimization algorithm. On the other hand, when γ is very large, then the attractiveness (and thus brightness) decreases dramatically, and all fireflies are short-sighted or equivalently fly in a deep foggy sky. This means that all fireflies move almost randomly, which corresponds to a random search technique.

In general, the FA corresponds to the situation between these two limit cases, and it is thus possible to fine-tune these parameters, so that FA can find the global optima as well as all the local optima simultaneously in a very effective manner. A further advantage of FA is that different fireflies will work almost independently, it is thus particular suitable for parallel implementation. It is even better than genetic algorithm and particle swarm optimization because fireflies aggregate more closely around each optimum. It can be anticipated that the interactions between different sub-regions are minimal in parallel implementation. Nowadays, mechanisms of firefly communication via luminescent flashes and their synchronization has been used effectively to solve the problems in various areas, such as in continuous constrained optimization [32], economic emissions load dispatch [3], image compression [18, 20], mixed variable structural optimisation [15], re-machining parameter optimization [43], scheduling [36], clustering [19, 37], parameter tuning [48], wireless network design [31], dynamic marketing pricing [22].

5.3 Benchmark Test Function

In this section, a benchmark test function (see Table 2) is selected from the literature to demonstrate the effectiveness of proposed TLBO and FA in dealing with function optimization problem. The chosen function is a non-linear minimization problem which has seven design variables and four non-linear inequality constraints.
Table 2

Benchmark test function (adapted from [35])

Benchmark test function

Minimize:

 
 

\( \begin{array}{*{20}c} {f\left( {\mathbf{x}} \right) \, = } \hfill & {\left( {x_{1} - 10} \right)^{2} \,+\, 5\left( {x_{2} - 12} \right)^{2} \,+\, x_{3}^{4} + 3\left( {x_{4} - 11} \right)^{2} } \hfill \\ {} \hfill & { + 10x_{5}^{6} + 7x_{6}^{2} + x_{7}^{4} - 4x_{6} x_{7} - 10x_{6} - 8x_{7} } \hfill \\ \end{array} \)

Subject to:

 
 

\( \begin{aligned} g_{1} \left( {\mathbf{x}} \right) &= - 127 + 2x_{1}^{2} + 3x_{2}^{4} + x_{3} + 4x_{4}^{2} + 5x_{5} \le 0 \hfill \\ g_{2} \left( {\mathbf{x}} \right) &= - 282 + 7x_{1} + 3x_{2} + 10x_{3}^{2} + x_{4} - x_{5} \le 0 \hfill \\ g_{3} \left( {\mathbf{x}} \right) &= - 196 + 23x_{1} + x_{2}^{2} + 6x_{6}^{2} - 8x_{7} \le 0 \hfill \\ g_{4} \left( {\mathbf{x}} \right) &= 4x_{1}^{2} + x_{2}^{2} - 3x_{1} x_{2} + 2x_{3}^{2} + 5x_{6} - 11x_{7} \le 0 \hfill \\ \end{aligned} \)

where:

 
 

\( - 10 \le x_{i} \le 10\,(i = 1,2, \ldots 7) \)

The parameter settings for TLBO and FA are as follows:
  • TLBO: Population size is 50, generations are 2,000, total number of function evaluations are 100,000;

  • FA: Population size is 20, generations are 5,000, total number of function evaluations are 100,000.

The numerical results obtained via FA and TLBO are outlined in Table 3.
Table 3

Comparison of results (10 runs) obtained by using FA and TLBO for Benchmark test function

 

Benchmark test function

 

FA

TLBO

f(x *)—best

680.7249

680.6305

f(x *)—worst

681.1270

680.6325

f(x *)—mean

680.8900

680.6318

x *—best

\( \left( {\begin{array}{*{20}c} {2.349317} \\ {1.948980} \\ { - 0.442630} \\ {4.366729} \\ { - 0.638468} \\ {1.027457} \\ {1.614927} \\ \end{array} } \right) \)

\( \left( {\begin{array}{*{20}c} {2.331588} \\ {1.951348} \\ { - 0.477926} \\ {4.365752} \\ { - 0.626203} \\ {1.032656} \\ {1.593086} \\ \end{array} } \right) \)

f(x *) Objective Function Value; x * Optimum Solution

The optimum solution is at x * = (2.330499, 1.951372, −0.4775414, 4.365726, −0.6244870, 1.1038131, 1.594227) with objective function value f(x *) = 680.6300573 [35]. Although, we only test ten runs on each method, it can be observed from Table 3, both FA and TLBO work fine on solving benchmark test function.

5.4 Benchmark Engineering Design Optimization Problem

In this section, a constrained benchmark engineering design problem (i.e., pressure vessels design) is selected from the literature to test the effectiveness of proposed TLBO and FA in dealing with constrained optimization problem. For a given volume and working pressure, the basic aim of designing a pressure vessel is to get the total cost minimized. The typical design variables are such as the thickness of the head and body, the inner radius, and the length of the cylindrical section. This is a well-known test problem for optimization and the standard form can be found in Table 4.
Table 4

Benchmark engineering design problem (adapted from [35])

Benchmark engineering design problem

Minimize

 
 

\( \begin{aligned} f(x)=&\;0.6224{d_{1}}rL+1.7781{d_{2}}{R^{2}}\\&+3.1661\,{D_{I}^{2}}L+19.84{d_{I}^{2}}r \end{aligned} \)

Subject to

 
 

\( \begin{aligned} g_{1} \left( {\mathbf{x}} \right) &= - d_{1} + 0.0193r \le 0 \hfill \\ g_{2} \left( {\mathbf{x}} \right) &= - d_{2} + 0.00954r \le 0 \hfill \\ g_{3} \left( {\mathbf{x}} \right) &= - \pi r^{2} L - \frac{4}{3}\pi r^{3} + 1296000 \le 0 \hfill \\ g_{4} \left( {\mathbf{x}} \right) &= L - 240 \le 0 \hfill \\ \end{aligned} \)

where

 
 

\( \begin{array}{*{20}c} {0 \le d_{1} \le 99} \hfill & {0 \le d_{2} \le 99} \hfill \\ {10 \le r \le 200} \hfill & {10 \le L \le 200} \hfill \\ \end{array} \)

To evaluate the performance of FA and TLBO algorithms for optimizing the design of pressure vessels, we also run the simulation 10 times under the same parameter settings as previously mentioned. The numerical results are shown in Table 5.
Table 5

Comparison of results (10 runs) obtained by using FA and TLBO for Benchmark engineering design problem

 

Benchmark engineering design problem

 

FA

TLBO

f(x *)—best

6117.2432

5885.3330

f(x *)—worst

6418.8358

5885.3890

f(x *)—mean

6263.9090

5885.3453

x *—Best

\( \left( {\begin{array}{*{20}l} {0.828478} \\ {0.425984} \\ {42.619529} \\ {172.301498} \\ \end{array} } \right) \)

\( \left( {\begin{array}{*{20}l} {0.778169} \\ {0.384649} \\ {40.319619} \\ {200} \\ \end{array} } \right) \)

f(x *) Objective Function Value; x * Optimum Solution

As it shown in Table 5, according to other similar studies such as [9], both solutions obtained through TLBO and FA are reasonable in which TLBO can generate a slightly better results than FA on the test problem.

From the above mentioned two examples, we can see that TLBO and FA are very good optimizer and suitable for many applications.

6 Experimental Study by Using TLBO and FA

Witnessing the capability of TLBO and FA, in this section, we decide to carry out an experimental study to solve our focal problem. Suppose that we have the following parameters and constraints (arrayed in Table 6) for our RFID system.
Table 6

Data used for RFID system

Stage

10 5 α i

β i

v i

w i

V

C

W

T

1

1.0

1.5

1

6

200

500

450

2000

2

2.3

1.5

2

6

3

0.3

1.5

3

8

4

2.3

1.5

2

7

Based on Eq. (1), we run FA and TLBO 10 times for each and the simulation results are listed in Tables 7 and 8, respectively.
Table 7

Convergence results (10 runs) obtained by using FA and TLBO

 

Convergence results of f(r, n)

 

FA

TLBO

f(rn)—best

99.6152

99.6156

f(rn)—worst

99.6118

99.6153

f(rn)—mean

99.6139

99.6155

Table 8

Best results (10 runs) obtained by using FA and TLBO for RFID system

 

RFID system

 

FA

TLBO

f ( r n )

99.6152

99.6156

n 1

5

5

n 2

5

5

n 3

4

4

n 4

5

5

r 1

72.0518

72.4646

r 2

74.0182

73.9552

r 3

82.1285

81.9053

r 4

74.8874

74.7276

From Table 8, we can see that in order to keep our RFID system reliability at the highest level, the components number and the corresponding reliability should be designed based on the results obtained via TLBO and FA.

7 Future Research Directions

Since our targeted question belongs to a class of constrained nonlinear mixed-integer programming problem which means the solution of this kind of problems consists two parts, i.e., a real part and an integer part. As the searching space and complexity of these two parts are different, it might be more promising to use different searching mechanisms to obtain individual optimal solution for each of these two parts. Therefore one possible future research direction is to employ two algorithms to search the real part and the integer part, respectively.

In order to maximize a system’s reliability, except the reliability-redundancy allocation solution, one can also consider other options such as enhancing the component reliability [30]. Since there are many other communication systems may utilize the similar frequencies within the communication range of RFID which in turn could interfere the reliability of the RFID system, a good filter design for RFID receiver is always a necessary. Therefore an immediate extension of the current research would be employing suitable CI methods to optimize the performance of signal filter component within a RFID reader.

8 Conclusions

Re-manufacturability classification based on radio frequency identification (RFID) system is a great concept transition and innovation. The idea is to “take the initiative to prevent problems”, which can greatly save resources and energy of the whole world and bring enormous economic benefits as well as social benefits. However, recently the growing interest in cyber physical re-manufacturability pre-evaluation faces major challenges due to the error prone nature of RFID devices. The focus of our work is complementary to the inherent unreliability of RFID systems, and ask whether the reliability can be improved using more redundant components (i.e., RFID readers) in wide range type series. In this chapter, we first formulate our focal scenario as a reliability-redundancy allocation problem (RRAP). Then, two of the recently developed computational intelligence approaches called teaching—learning-based optimization (TLBO) which is based on the effect of the influence of a teacher on the output of learners in a class, and firefly algorithm (FA) which is based on the social (flashing) behaviour of fireflies, or lighting bugs, in the summer sky in the tropical temperature regions, are employed to address our focal problem. Simulation results suggest that the proposed TLBO and FA are viable optimization techniques in improving the RFID classification system’s reliability.

References

  1. 1.
    M. AliIlgin, S.M. Gupta, Performance improvement potential of sensor embedded products in environmental supply chains. Resour. Conserv. Recycl. 55(6), 580–592 (2011).Google Scholar
  2. 2.
    T. Amezquita, R. Hammond, M. Salazar, B. Bras, Characterizing the re-manufacturability of engineering systems. Paper presented at the ASME advances in design automation conference, Boston, Massachusetts, USA, pp. 271–278, 17–20 Sept 1995Google Scholar
  3. 3.
    T. Apostolopoulos, A. Vlachos, Application of the firefly algorithm for solving the economic emissions load dispatch problem. Int. J.Combinatorics, Article ID 523806, 1–23 (2011)CrossRefGoogle Scholar
  4. 4.
    N. Aras, T. Boyaci, V. Verter, The effect of categorizing returned products in re-manufacturing. IIE Trans. 36, 319–331 (2004)CrossRefGoogle Scholar
  5. 5.
    B.G. Babu, M. Kannan, Lightning bugs. Resonance 7(9), 49–55 (2002)CrossRefGoogle Scholar
  6. 6.
    H.-G. Beyer, H.-P. Schwefel, Evolution strategies: a comprehensive introduction. J. Nat Comput. 1(1), 3–52 (2002)CrossRefMATHMathSciNetGoogle Scholar
  7. 7.
    J.D. Blackburn, V.D.R. Guide, G.C. Souza, L.N.V. Wassenhove, Reverse supply chains for commercial returns. Calif. Manag. Rev. 46(2), 6–22 (2004)CrossRefGoogle Scholar
  8. 8.
    B. Bras, M.W. McIntosh, Product, process, and organizational design for re-manufacture—an overview of research. Rob. Comput. Integr. Manuf. 15, 167–178 (1999)CrossRefGoogle Scholar
  9. 9.
    L.C. Cagnina, S.C. Esquivel, C.A. Coello, Solving engineering optimization problems with the simple constrained particle swarm optimizer. Informatica 32, 319–326 (2008)MATHGoogle Scholar
  10. 10.
    D.J. Cook, J.C. Augusto, V.R. Jakkula, Ambient intelligence: technologies, applications, and opportunities. Pervasive Mob. Comput. 5, 277–298 (2009)CrossRefGoogle Scholar
  11. 11.
    M. Črepinšek, S.-H. Liu, L. Mernik, A note on teaching–learning-based optimization algorithm. Inf. Sci. 212, 79–93 (2012)CrossRefGoogle Scholar
  12. 12.
    Y. Du, H. Cao, F. Liu, C. Li, X. Chen, An integrated method for evaluating the re-manufacturability of used machine tool. J. Clean. Prod. 20, 82–91 (2012)CrossRefGoogle Scholar
  13. 13.
    M. Fleischmann, H.R. Krikke, R. Dekker, S.D.P. Flapper, A characterisation of logistics networks for product recovery. OMEGA 28, 653–666 (2000)CrossRefGoogle Scholar
  14. 14.
    M. Fleischmann, JAEEv Nunen, B. Gräve, Integrating closed-loop supply chains and spare-parts management at IBM. Interfaces 33(6), 44–56 (2003)CrossRefGoogle Scholar
  15. 15.
    A.H. Gandomi, X.-S. Yang, A.H. Alavi, Mixed variable structural optimization using firefly algorithm. Comput. Struct. 89, 2325–2336 (2011)CrossRefGoogle Scholar
  16. 16.
    V.D.R. Guide, R.H. Teunter, L.N.V. Wassenhove, Matching demand supply to maximize profits from re-manufacturing. Manuf. Serv. Oper. Manage. 5(4), 303–316 (2003)Google Scholar
  17. 17.
    R. Hammond, T. Amezquita, B. Bras, Issues in the automotive parts re-manufacturing industry-a discussion of results from surveys performed among re-manufacturers. Int. J. Eng. Des. Autom.—Spec Issue Environmentally Conscious Des. Manufact. 4(1), 27–46 (1998)Google Scholar
  18. 18.
    M.-H. Horng, Vector quantization using the firefly algorithm for image compression. Expert. Syst. Appl. 39, 1078–1091 (2012)CrossRefGoogle Scholar
  19. 19.
    M.-H. Horng, Y.-X. Lee, M.-C. Lee, R.-J. Liou, Firefly meta-heuristic algorithm for training the radia basis function network for data classification and disease diagnosis. in ed. by R. Parpinelli Theory and New Applications of Swarm Intelligence, Chapter 7, pp. 115–132: In-Tech (2012)Google Scholar
  20. 20.
    M.-H. Horng, R.-J. Liou, Multilevel minimum cross entropy threshold selection based on the firefly algorithm. Expert. Syst. Appl. 38, 14805–14811 (2011)CrossRefGoogle Scholar
  21. 21.
    Y.-C. Hsieh, P.-S. You, An effective immune based two-phase approach for the optimal reliability. Appl. Math. Comput. 218, 1297–1307 (2011)CrossRefMathSciNetGoogle Scholar
  22. 22.
    J. Jumadinova, P. Dasgupta, Firefly-inspired synchronization for improved dynamic pricing in online markets. Paper presented at the 2nd IEEE international conference on self-adaptive and self-organizing Systems, pp. 402–412, (2008)Google Scholar
  23. 23.
    Ö. Karaer, H.L. Lee, Managing the reverse channel with RFID-enabled negative demand information. Prod. Oper. Manage. 16(5), 625–645 (2007)CrossRefGoogle Scholar
  24. 24.
    M. Kärkkäinen, T. Ala-Risku, K. Främling, The product centric approach: a solution to supply network information management problems? Comput. Ind. 52(2), 147–159 (2003)CrossRefGoogle Scholar
  25. 25.
    M. Kärkkäinen, T. Ala-Risku, J. Holmström, Increasing customer value and decreasing distribution costs with merge-in-transit. Int. J. Phys. Distrib. Logistics. Manage. 33(2), 132–148 (2003)CrossRefGoogle Scholar
  26. 26.
    J. Kennedy, R. C. Eberhart, Particle swarm optimization. Paper presented at the IEEE International Joint conference on neural networks (1995)Google Scholar
  27. 27.
    H. Krikke, I. l. Blanc, S. v. d. Velde, Product modularity and the design of closed-loop supply chains. Calif. Manage. Rev. 46(2), 23–38 (2004)Google Scholar
  28. 28.
    A. Kulkarni, D. Ralph, D. McFarlane, Value of RFID in re-manufacturing. Int. J. Serv. Oper. Inf. 2(3), 225–252 (2007)Google Scholar
  29. 29.
    A. G. Kulkarni, A. K. N. Parlikad, D. C. McFarlane, M. Harrison, Networked RFID systems in product recovery management. Paper presented at the IEEE international symposium on electronics and the environment (ISEE 2005), pp. 66–71 (2005)Google Scholar
  30. 30.
    W. Kuo, V.R. Prasad, An annotated overview of system-reliability optimization. IEEE Trans. Reliab. 49, 176–187 (2000)CrossRefGoogle Scholar
  31. 31.
    R. Leidenfrost, W. Elmenreich, Establishing wireless time-triggered communication using firefly clock synchronization approach. Paper presented at the 2008 international workshop on intelligent solutions in embedded systems, pp. 1–8, (2008)Google Scholar
  32. 32.
    S. Łukasik, S. Żak, Firefly algorithm for continuous constrained optimization tasks Computational Collective Intelligence. Semantic Web, Social Networks and Multiagent Systems, LNCS 5796. (Berlin, Spinger-Verlag, 2009), pp. 97–106Google Scholar
  33. 33.
    Z. Michalewicz, Genetic Algorithms + Data Structures = Evolution Programs, 3rd edn. (Springer-Verlag, Berlin Heidelberg, 1996)MATHGoogle Scholar
  34. 34.
    J. A. E. E. v. Nunen, R. Zuidwijk, E-enabled closed-loop supply chains. Calif. Manage. Rev. 46(2), 40–54, (2004)Google Scholar
  35. 35.
    R.V. Rao, V.J. Savsani, D.P. Vakharia, Teaching–learning-based optimization: a novel method for constrained mechanical design optimization problems. Comput. Aided. Des. 43, 303–315 (2011)CrossRefGoogle Scholar
  36. 36.
    M.K. Sayadi, R. Ramezanian, N. Ghaffari-Nasab, A discrete firefly meta-heuristic with local search for makespan minimization in permutation flow shop scheduling problems. Int. J. Ind. Eng. Comput. 1, 1–10 (2010)CrossRefGoogle Scholar
  37. 37.
    J. Senthilnath, S.N. Omkar, V. Mani, Clustering using firefly algorithm: performance study. Swarm Evol. Comput. 1, 164–171 (2011)CrossRefGoogle Scholar
  38. 38.
    L. Shu, W. Flowers, A structured approach to design for re-manufacture. Intell. Concurrent Des.: Fundam. Methodol. Model. Pract. 66, 13–19 (1993)Google Scholar
  39. 39.
    R. Storn, K. Price, Differential evolution—a simple and efficient heuristic for global optimization over continuous spaces. J. Global. Optim. 11(4), 341–359 (1997)CrossRefMATHMathSciNetGoogle Scholar
  40. 40.
    E. Sundin, B. Bras, Making functional sales environmentally and economically beneficial through product re-manufacturing. J. Clean. Prod. 13, 913–925 (2005)CrossRefGoogle Scholar
  41. 41.
    L. Wang, L.-P. Li, A coevolutionary differential evolution with harmony search for reliability–redundancy optimization. Expert. Syst. Appl. 39, 5271–5278 (2012)CrossRefGoogle Scholar
  42. 42.
    X. Wu, Research on design management based on green re-manufacturing engineering. Syst. Eng. Procedia 4, 448–454 (2012)CrossRefGoogle Scholar
  43. 43.
    B. Xing, W.-J. Gao, Computational intelligence in re-manufacturing. 701 E. Chocolate Avenue, Suite 200, Hershey PA 17033: IGI Global, ISBN 978-1-4666-4908-8 (2014)Google Scholar
  44. 44.
    B. Xing, W.-J. Gao, T. Marwala, The applications of computational intelligence in radio frequency identification research. Paper presented at the IEEE international conference on systems, man, and cybernetics (IEEE SMC), 14–17 Oct, (Seoul, Korea, 2012) pp. 2067–2072Google Scholar
  45. 45.
    X.-S. Yang, Nature-inspired metaheuristic algorithms 2nd edn. (UK, Luniver Press) ISBN 978-1-905986-28-6 (2008)Google Scholar
  46. 46.
    X.-S. Yang, Firefly algorithms for multimodal optimization, in SAGA 2009, LNCS 5792, ed. by O. Watanabe, T. Zeugmann (Springer-Verlag, Berlin Heidelberg, 2009), pp. 169–178Google Scholar
  47. 47.
    X.-S. Yang, Firefly algorithm, stochastic test functions and design optimisation. Int. J. Bio-Inspired Comput. 2(2), 78–84 (2010)CrossRefGoogle Scholar
  48. 48.
    X.-S. Yang, Chaos-enhanced firefly algorithm with automatic parameter tuning. Int. J. Swarm Intell. Res. 2(4), 1–11 (2011)CrossRefGoogle Scholar
  49. 49.
    J. Yao, S. Zhu, The research of design system for re-manufacturing. N. Technol. N. Process 5, 22–24 (2004)Google Scholar
  50. 50.
    W.-C. Yeh, T.-J. Hsieh, Solving reliability redundancy allocation problems using an artificial bee colony algorithm. Comput. Oper. Res. 38, 1465–1473 (2011)CrossRefMathSciNetGoogle Scholar
  51. 51.
    H. Yüksel, Design of automobile engines for re-manufacture with quality function deployment. Int. J. Sustain. Eng. 3(3), 170–180 (2010)CrossRefGoogle Scholar
  52. 52.
    Z.-N. Zhang, Z.-L. Liu, Y. Chen, Y.-B. Xie, Knowledge flow in engineering design: an ontological framework. Proc. Ins. Mech. Eng. Part C: J. Mech. Eng. Sci. 227(4), 760–770 (2013)CrossRefGoogle Scholar
  53. 53.
    C. Zikopoulos, G. Tagaras, On the attractiveness of sorting before disassembly in re-manufacturing. IIE Trans. 40, 313–323 (2008)CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media Singapore 2014

Authors and Affiliations

  1. 1.Department of Mechanical and Aeronautical Engineering, Faculty of Engineering, Built Environment and Information TechnologyUniversity of PretoriaPretoriaSouth Africa

Personalised recommendations