Professional Documents
Culture Documents
Mehran Tamjidy , Shahla Paslar , B.T. Hang Tuah Baharudin , Tang Sai Hong & M.K.A.
a
Ariffin
a
To cite this article: Mehran Tamjidy, Shahla Paslar, B.T. Hang Tuah Baharudin, Tang Sai Hong & M.K.A. Ariffin (2014):
Biogeography based optimization (BBO) algorithm to minimise non-productive time during hole-making process, International
Journal of Production Research, DOI: 10.1080/00207543.2014.965356
To link to this article: http://dx.doi.org/10.1080/00207543.2014.965356
1. Introduction
Hole-making is a type of machining processes that are specically used to cut a hole into a part. Drilling, tapping, reaming and punching are typical hole making operations and they compose a large portion of machining processes for various industries such as electronic and plastic injection mould (Ghaiebi and Solimanpur 2007). In some cases, a plastic
injection mould could have more than 100 holes with dissimilar diameters, depths, tolerance, and different surface conditions, in presence of various tool requirements and a large number of tool switches (Kolahan and Liang 2000). In real
production systems, some small holes can be executed to its desired size with only one tool, or by using a sequence of
tools with different diameters. Moreover, a tool can be used to perform several holes in different position as the nal or
intermediate operation (Liu et al. 2013). In fact, most of hole-making problems are NP-hard.
The costs of hole-making process is directly associated with the total production time that consists of travelling time,
the time required to move the tool between two holes, switching time, the time to change the tool for next operation,
cutting time, and the time to execute a hole (Onwubolu and Clerc 2004). According to the Merchant (1985) report, on
average 70% of the total time takes by the part and tool movements in the manufacturing process. One of the crucial
issues in current mass production industry is to minimise non-productive time such as travelling and switching time by
means of automated machining.
Due the aforementioned complexities of hole-making problem metaheuristic approaches have found relatively more
favour in literature to optimise the hole-making process. Kolahan and Liang (2000) employed a tabu-search approach to
minimise the total processing cost for hole-making operations. They considered four issues, including tool travel time,
tool switching time, tool selection and machining speed specication and used the tabu-search algorithm to nd the
solution. Onwubolu and Clerc (2004) proposed a new heuristic approach based on particle swarm optimization (PSO) to
optimise the operation path of automated or computer numerical control (CNC) drilling process. In their study, rstly,
the tool path of a CNC drilling machine has been modelled as a travelling salesman problem (TSP); then a model for
approximate prediction of drilling time is developed. Finally, an adaptive PSO algorithm is used to solve the TSP.
Abbas, Aly, and Hamza (2010) developed an approach based on ant colony optimization (ACO) to solve the CNC drilling tool path optimization problem as TSP in special case of production with large number of holes. Oysu and Bingul
(2009) addressed three algorithms as approached of genetic algorithm (GA), simulate annealing (SA) and hybrid GASA
to minimise the tool path on three-axis milling robot on wood materials. Their experiments were compared between
these algorithms based on minimum airtime. Liu et al. (2013) presented a new method to optimise the process planning
M. Tamjidy et al.
of hole-making with several operations by means of different tool sets. Their optimization objectives were to minimise
the total time of the airtime and the tool switching time by aid of ACO algorithm. Medina-Rodrguez et al. (2012)
employed a parallel ACO algorithm to nd the best sequence of hole-making to create G code programing for the shortest cutting path. Ghaiebi and Solimanpur (2007) proposed an ant algorithm to deal with the optimization of hole-making
operations in which a hole can be completed by using several tools. Moreover, a 01 nonlinear mathematical model is
formulated to minimise the summation of tool travel time and switch time.
The main contribution of this study is to solve the hole-making problem by applying an efcient metaheuristic algorithm to minimise tool travel time and switch time. In 2008, a new population-based evolutionary algorithm based on
geographic distribution of biological organisms was rstly introduced by Simon (2008), entitled biogeography based
optimization (BBO). He stated that this method is novel method for solving the NP hard problem. Although BBO is a
naturally inspired algorithm, it has some fundamental distinction from common natural algorithms such as GA, PSO or
ACO. In the BBO, the initial population is not discarded among different generation. Instead, the migration concept is
used to modify the population. As another distinction, in each generation, the tness function is not used directly to
modify the population in which BBO uses tness to determine the immigration and emigration rates.
Regarding the application of BBO algorithm in manufacturing system, BBO has been employed to solve scheduling
problem (Rahmati and Zandieh 2012; Attar, Mohammadi, Tavakkoli-Moghaddam 2013) and to design optimal placement of phasor measurement units (Jamuna and Swarup 2012). According to their comparative study of BBO with other
popular metaheuristic algorithms; they concluded that BBO is capable to obtain the better results in terms of solution
quality and convergence characteristics.
Herein we propose a linear mathematical model for hole-making problem where the machining process of a hole
consists of several individual operations with various machining tools and different constraints such as precedence and
assignment are imposed on the process. This model is developed based on the models presented by Ghaiebi and
Solimanpur (2007) and Liu et al. (2013). Since this problem is considered to be NP-hard, a BBO algorithm is employed
to nd the optimum/near optimum solution within short amount of time. Our algorithms are implemented to minimise
the summation of two objective functions, namely, tool travelling time and tool switching time. The performance of the
proposed algorithm is compared with some common meta-heuristic algorithms which have been widely used in the literature such as GA and ACO reported in Liu et al. (2013) and Ghaiebi and Solimanpur (2007). Rest of this paper is
organised as follows: Section 2 describes the problem environment, Section 3 presents the mathematical model, Section 4
develops the proposed BBO algorithm, Section 5 evaluates the performance of proposed BBO algorithm, and Section 6
provides the nal conclusion.
2. Problem environment
In hole-making process, some small holes can be machined to their nal size through only one operation, but most of
the holes need to be machined by several tools with different diameters, which is inevitable when the hole-diameter is
relatively large. In this case, a pilot hole can be drilled by means of a smaller tool and then enlarge the hole to its
desired size by using a tool of larger diameter, probably completed by reaming or tapping when needed.
This paper considers the hole-making problem that includes a set of holes i 2 I that each hole is supposed to be
accomplished by a number of tools t 2 T . Each tool-hole combination is considered as an operation j 2 ni . Each operation is supposed to be placed in a position p 2 N (N = total number of operation) in a sequence.
Drilling machine has a worktable can move in two directions x and y. Measuring the distance between two sequential holes is based on how the worktable moves. The movement of the two-axis drill machine can be in both direction
simultaneously or just in one direction at each time (Ghaiebi and Solimanpur 2007). The mentioned worktable movements can be measured by the Euclidean and rectilinear distance function. In this study both distance functions, Euclidean and rectilinear, are considered. For two adjacent hole-operation positions, dii0 denotes the distance between two
different holes i and i0 located at coordinates xi ; yi and xi0 ; yi0 respectively which executed by same tool. The distance between hole i and hole i0 can be expressed as follows.
(a) Euclidean distance
q
0
(1)
dii 2 xi xi0 2 yi yi0 2
(b) Rectilinear distance
dii0 jxi xi0 j jyi yi0 j
(2)
By considering the speed of worktable in x direction as vx and y direction as vy in case of rectilinear distance and v in
Euclidean distance, the travelling time between two sequential hole i and hole i0 , denoted by TTii0 , can be calculated by
Equations (3) and (4).
(a) Euclidean distance
q
2
xi xi0 2 yi yi0 2
(3)
TTii0
v
(b) Rectilinear distance
TTii0
vx
vy
(4)
Hole-making decision for each tool-hole combination is made by considering some assumptions, which are made as
follow:
(1) Each hole can be completed in multi-passes if there is particular tool sequence required for completing the
hole.
(2) When a particular tool completes two sequential operations the tool switching time will be zero (0).
(3) In this study the start point of spindle is located in point (0, 0).
(4) Tool life of each tool type is adequate for executing entire holes assigned to it.
(5) The model considers based on both Euclidean and rectilinear distance during the movement of spindle from
one hole to the next.
The subsequent example shows the simple part with 3 holes and their required number of tools in order to complete
these holes. Figure 1(a) depicts the example part with their required number of tools. Based on the hole-operation information given in Table 1, initially the rst operation of each hole must be selected for machining. In this case, rst, one
of the operations o11 ; o21 and o31 must be executed before their successive operations if there is any. At each stage of
hole-making the sequence of hole-operations must be maintained. For better visualisation Figure 1(b) shows a technical
sequence of operations for each hole and a possible sequence of tool travel path.
The aim of this study is to optimise the tool-hole sequence by employing BBO algorithm in order to reduce the
non-productive time including tool travelling time and tool switching time. A tool switching time can be dened as
summation of tool to tool changing time and a tool travelling time that occurs due to the tool trip from the executed
hole to the tool magazine and then trip from the tool magazine to the next hole.
3. Model formulation
In this section, a mathematical model is presented to clearly specify the key parameters and their inuence on the holemaking problem. The present model modies the model presented in Ghaiebi and Solimanpur (2007) and Liu et al.
(2013).
Figure 1. (a) Example part of tool sequence for hole-making. (b) Technical sequence of operations.
M. Tamjidy et al.
Hole 1
Hole 2
Hole 3
t1
t1 ; t2
t1 ; t2 ; t3
o11
o21 o22
o31 o32 o33
3.1 Notation
Subscripts
i; i0
j; j0
t; t 0
p; p0
hole indices; 1 i; i0 I
operation indices; 1 j; j0 ni
tool indices; 1 t; t 0 T
position in the sequence; 1 p; p0 N
Parameters
TTii0
TToi
TTiM
N
TS
and sets
travelling time between hole i and hole i0
travelling time between starting point and hole i
travelling time between hole i and coordination of tool magazine
total number of positions in a sequence and or total number of operations
tool to tool changing time
Decision variables
Lijtpi0 j0 t0 p1 equal to one if the spindle moves between two adjacent position of sequence to process operation j and j0
of hole i and i0 by tool t and t 0 respectively; zeros otherwise
Xijtp
equal to one if operation j of hole i is processed by using tool t on pth position of the sequence; zero
otherwise
3.2 Mathematical model
The mathematical model which is able to take into account an objective function and various constraints arising from
the problem environment can be stated as follows.
Min
N 1
X X XXX X X
i0
0
i 6X
i
XX
i
j0
t0
0
t t
Lijtpi0 j0 t0 p1 TTii0
N 1
XXXXX X X
i0
p1
j0
t0
0
t 6 t
Xijt1 TToi
(5)
Subject to:
X
Xijtp 1
8i; j; t 2 ComTij
(6)
Xijtp 1 8p
(7)
XX X
i
t2ComTij
Xijtp
N
X
Xij1tp0
8i; j 6 ni ; t 2 ComTij
(8)
p0 p1
(9)
8i; j; t; p
(10)
(11)
Lijtpi0 j0 t0 p1 2 f0; 1g
8i; i0 ; j; j0 ; t 2 ComTij ; t 0 2 ComTi0 j0 ; t 6 t 0 ; p 2 f1; 2; . . .; N 1g or i 6 i0 ; t t 0
(12)
Equation (5) prescribes the objective function that minimise the summation of the tool travelling time between holes,
tool switching time and tool travelling time of rst position respectively. Tool travelling time can be calculated by summation of tool travelling between different holes in the adjacent position that processed with the same tool. The second
component in objective function, tool switching time, is formulated by total summation of tool to tool changing time
and tool travelling time, occurs when spindle has to move to tool magazine to change the tool for the next operation.
The tool changing point is located at zero in x-axis and value of y-axis coordination of last hole i which is executed by
tool t. The last component represents the rst tool travelling time between starting point with coordination value of
(0, 0) and hole in rst position of sequence.
Constraint (6) indicates that for each tool-hole operation only one position of the sequence can be assigned. Constraint (7) ensures that for each position in the sequence only one tool-hole operation can be occupied. Constraint (8)
prescribes the precedence constrain between operations of each hole. Constraints (9) and (10) state the relationship of
two consecutive position of sequence. The result of Equations (9) and (10) should be equal to one if the mentioned
positions have been processed by two different tools or when two different holes processed with same tool, otherwise is
zero. Constraints (11) and (12) restrict the decision variables into zero-one values.
4. Proposed algorithms
4.1 BBO algorithm
The BBO is a new evolutionary algorithm among the popular meta-heuristic approaches which have arisen as attractive
optimization algorithms due to their competitive results (Simon 2008). This population-based algorithm is a naturally
inspired algorithm in which mimics the migration process of species for solving engineering problems (Rahmati and
Zandieh 2012). This algorithm has revealed notable performance on many well-known case studies (Dawei, Simon, and
Ergezer 2009). BBO algorithm starts the optimization process with a number of candidate solutions, called habitats or
islands. Each island feature is considered by a suitability index variable (SIV). Each habitat is characterised by a quantitative performance index, named habitat suitability index (HSI).
The main principal of BBO is based on immigration and emigration of species in a habitat, known as migration. With probabilistic migration, BBO is able to share more information from good solutions to poor ones. In
other words, this algorithm prevents the good solutions to be demolished during the evolution. Thus, it can efciently utilise the characteristics and information of population in per iteration. Without mutation operator that can
increase the diversity among the population, the solutions with high HSI have tendency to be more dominant in
population.
M. Tamjidy et al.
Hj SIV
(13)
The immigration and emigration rates are functions of the solutions tness. They can be evaluated by Equations
(14) and (15) respectively.
ki
ki I 1
n
li E
ki
n
(14)
(15)
In Equations (14) and (15), I and E represent the maximum possible immigration and emigration rate respectively;
ki is the rank of habitat i after sorting all habitats according to their HSI; and n is the number of solutions in the population. It is clear that the better solution has higher emigration and lower immigration rate, while the converse is true for
a poor solution. Often, I and E set equal to one or slightly less than one (Ma and Simon 2011).
After determining the immigrating and emigrating habitats, the migration process can be performed like crossover in
evolutionary algorithms. In this study, to do the migration, improved precedence operation crossover (IPOX) is adopted
from Zhang et al. (2007) for operation sequence. This effective migration operator is described as follows.
Step 1: Divide the set of holes f1; 2; 3; . . .ng, into two non-empty groups H1 and H2 randomly.
Step 2: Direct copies (same position) those numbers in H1 from immigrating habitat to the modied habitat.
Step 3: Indirect copies (same order) those numbers in H2 from emigrating habitat to the modied habitat.
Figure 3. IPOX migration operator for tool-hole operation sequence for H1: [3] and H2: [1, 2].
Figure 3 illustrates an example of the IPOX of three holes f1; 2; 3g with 1, 2 and 3 operations respectively. The set
of hole is divided into two sets H1 f3g and H2 f1; 2g randomly. The migration of immigrating habitat and emigrating habitat generates the modied habitat f3; 2; 1; 3; 2; 3g. It can be seen that modied habitat preserves the position
and order of hole f3g in immigrating habitat and the order of hole f1; 2g in emigrating habitat respectively. Therefore
IPOX is excellent in the characteristics-preservingness.
4.1.4 Mutation
In BBO, mutation rate is inversely related to the solution probability and can be calculated by Equation (16).
Pi
mi mmax 1
Pmax
(16)
Figure 4. Example of mutation operator for tool-hole operation sequence, (a) swap, (b) reversion, (c) insertion.
M. Tamjidy et al.
Based on aforementioned BBO Operators, the BBO algorithm to solve the hole-making problem is described as follows.
Find
;
Ftot
ind 1; 2; . . .; N
(17)
where Psel is the probability of choosing the indth individual, N is the population size, Find is the indth individual tness, and Ftot is the total tness of all individuals in the current generation.
Figure 5. (a) The method to determine the position of the ith hole in the workpiece. (b) The position of 10 holes in the workpiece.
5. Computational results
In order to examine the performance and effectiveness of proposed BBO algorithm, we use two test problems to conduct computational tests and comparison on both small-scale and large-scale instances for hole-making process. Test
problem 1 addresses to the example adopted from Liu et al. (2013). Test Problem 2 corresponds the second example
addresses by Ghaiebi and Solimanpur (2007). The mentioned test problems are described in the following sections.
These test problems are implemented by the proposed BBO algorithm coded using Matlab R2013a and run on
Intel core 2 Duo CPU T8100 at 2.1 GHZ, 3 GB RAM computer with windows 7.
The BBO parameters are set in this study after number of careful runs as follows; the habitat size (N) = 100, maximum migration and immigration rate of each habitat = 1 and mutation probability = 0.3.
5.1 Test problem 1
Test problem 1 considers a part with 42 (6 7) holes in which different number of tools is employed to complete the
desired size of each hole. The data about the distances between the holes, the diameter and type of holes are shown in
Figure 6(a). Table 2 presents a short description of these tool-hole combinations. This test problem corresponds to the
example addressed by Liu et al. (2013). The objective of test problem 1 is to minimise the total non-productive time
m
(auxiliary time). In this example the speed of spindle v and tool to tool changing time TS are set to 12 min
and 5 s
respectively. The tool magazine is located beside the column, so the tool changing point will move along the line parallel to y-axis, where x Xc . In this paper, it is assumed that Xc 0. Two measuring distance methods, Euclidian and
rectilinear, are used to test the performance of proposed BBO algorithm in the following sections. The proposed BBO
algorithm was applied to determine sequences of operations where a hole consists of several individual operations with
different machining tools.
M. Tamjidy et al.
10
11
Number of holes
Tool sequence
18
10
14
26
2345
18
Bh
Th
Sh
adopted from Ghaiebi and Solimanpur (2007). Each city in TSP indicates the position of each hole that must be drilled
and visited once (only one operation for each hole), and the tool will return to the initial point after completing its tour.
This test problem consists of 6 test instances, 5, 10, 15, 20, 25 and 50 holes respectively. To make the attempted problems reproducible, the arrangement of holes in the tested problems is considered as follows:
p
The number of rows in the workpiece is b I c where I is the total number of holes.
The centre to centre distance of holes in each direction is assumed to be 2 cm.
The position of the ith hole is determined as shown in Figure 5(a).
For example, the position
p of holes in the workpiece when I 10 is shown in Figure 5(b). As seen in this gure,
the number of rows is b 10c 3. The objective of test problem 2 is to minimise the tool path.
5.3 Performance of proposed BBO algorithm for test problem 1
Computational performance of proposed BBO algorithm is evaluated by considering the test problem 1. Table 3
describes a comparison between solution quality among the best near optimal solutions acquired by proposed BBO algorithm, proposed GA and the results of ACO proposed by Liu et al. (2013) for ten times computation when considering
non-productive time (auxiliary time) as objective function for two different measuring distance methods, Euclidian and
rectilinear. The rst column indicates the measuring distance methods name. In the second column, the best results of
proposed BBO algorithm are shown. The remaining columns reports the best results of the two algorithms, proposed
GA and ACO proposed by Liu et al. (2013), we compare with, together with the relative deviation with respect to our
BBO algorithm. Relative deviation criterion is used to compare the results of proposed BBO with the results obtained
by the two mentioned algorithms in term of solution quality. Relative deviation is obtained as follows.
dev
NPT comp NPT BBO
100
NPT comp
(18)
where NPT BBO is the best non-productive time obtained by our proposed BBO and NPT comp is the best non-productive time of the solution methodology that we compared ours to.
The results of this table shows how this simple version of BBO algorithm has signicant improvement to reach suboptimal solution in comparison with GA and the results reported by Liu et al. (2013). As can be seen from the results,
the proposed algorithm indicates 2.35 and 9.28% improvements in solution quality of proposed GA and ACO proposed
by Liu et al. (2013) respectively for Euclidian distance and 0.36 and 6.26% respectively for Rectilinear distance. The
overall result of the proposed BBO algorithm reveals an improvement in solution quality, 1.35 and 7.77% in average
with respect to the two solution approaches, proposed GA and ACO proposed by Liu et al. (2013) respectively. It can
be stated that this improvement may be obtained due to the ability of proposed BBO algorithm to efciently explore the
search space based on two main operators, migration and mutation. It is worth mentioning that the main difference of
proposed BBO algorithm with other evolutionary algorithms is in its selection strategies, one for migration and one for
mutation. Details of these selection strategies are previously explained in Sections 4.1.3 and 4.1.4.
To better understanding of the results found by proposed BBO algorithm, this example has been considered with
more details. Results that were reached by BBO algorithm for this test problem are shown in Table 4 as well as
Table 3. Computational results for test problem 1.
Measuring distance
Proposed BBO
Proposed GA
dev (%)
dev (%)
Euclidian distance
Rectilinear distance
119.2
136.85
122.07
137.35
2.35
0.36
131.4
146
9.28
6.26
Result of BBO
119.2
89
30
234518-6
Variables
Non-productive time
Tool travelling time
Tool switching time
Tool sequencing
122.07
92.07
30
2316845
Result of GA
131.4
101.4
30
2345618
Euclidian distance
136.85
106.85
30
2345186
Result of BBO
137.35
107.35
30
2345186
146
116
30
2345618
Rectilinear distance
Result of GA
12
M. Tamjidy et al.
13
Figures 6 and 7. Table 4 provides a detailed comparison between the results acquired by BBO algorithm, GA and ACO
model proposed by Liu et al. (2013) for this test problem. Figures 6 and 7 illustrate the optimal tool path represented
14
M. Tamjidy et al.
5
10
15
20
25
50
Proposed BBO
Proposed GA
dev (%)
dev (%)
12
24
32
40
52
104
12
24
32
40
52
112
0
0
0
0
0
7.14
12
24
32
40
52
136
0
0
0
0
0
23.53
by BBO algorithm in case of Euclidian and rectilinear distance measuring respectively. These diagrams depict the toolhole operations and tool sequences.
According to the results of Table 4, the solution obtained by proposed BBO algorithm is better than the two other
algorithms GA and ACO reported by Liu et al. (2013), since tool travelling time decreased from 101.4 and 92.07 to 89.2
for Euclidian distance and decreased from 116 and 107.35 to 106.85 for Rectilinear distance when the non-productive
time considered as objective function.
5.4 Performance of proposed BBO algorithm for test problem 2
In order to make a further investigation on the performance and effectiveness of proposed BBO algorithm, we ran the
BBO and GA on more challenging instances, test problem 2, that range from small to large scale instances. The results
of our algorithm are compared with ACO algorithm proposed by Ghaiebi and Solimanpur (2007) which is initially has
been compared with the dynamic programming (DP) and obtained optimum/near optimum solution in reasonable
amount of time for all test instances. As reported in their study, DP method is not able to solve the problems for 25 and
50 holes in a reasonable time due to the size of memory required for running DP. Table 5 describes a comparison
between solution quality among the best near optimal solutions acquired by proposed BBO algorithm, proposed GA and
the results of ACO proposed by Ghaiebi and Solimanpur (2007) for 25 trials when considering minimum tool path as
objective function. The rst column indicates the number of holes for each test instance. In the second column, the best
results of proposed BBO algorithm are shown. The remaining columns reports the best results of the two algorithms,
proposed GA and ACO proposed by Ghaiebi and Solimanpur (2007) that we compare with, together with the relative
deviation with respect to our BBO algorithm. Results presented in Table 5 can be analysed from the point of view of
different number of holes and their associated computational difculty. As can be seen, the proposed BBO performs
well by augment in the number of holes. The proposed algorithm indicates 23.53 and 7.14% improvements in solution
15
quality of proposed GA and ACO proposed by Ghaiebi and Solimanpur (2007) respectively for test instance with 50
holes. The results of other test instances are equal to the best results obtained by proposed GA and ACO proposed by
Ghaiebi and Solimanpur (2007). The overall result of the proposed BBO algorithm reveals an improvement in solution
quality, 1.19 and 3.92% in average with respect to the two solution approaches, proposed GA and ACO proposed by
Ghaiebi and Solimanpur (2007) respectively. For visualisation, Figure 8 depicts the optimum tool travel path of test
instance 6 in which 50 holes must be drilled in a part.
6. Conclusion
This research work developed a mathematical programming model for hole-making problem. Whereas, this problem is
considered as NP-hard, a newly meta-heuristic optimization algorithm, called BBO, is developed to solve the problem
of minimising the non-productive time including tool travelling time and tool switching time. The hole-making problem
in this study is concerned with the sequencing of various tool-hole combinations. The proposed solution methodology
has been tested on various test problems and the results obtained are compared with those from some of the existing
metaheuristic algorithms. As can be seen from this comparative study, it has been observed that the proposed algorithm
offers better results for majority of the test problems. However, application of the proposed BBO algorithm is limited to
certain instances where there is a part with different number of holes which are executed by one or more tool types.
This research can be further extended to see how the performance of BBO algorithm can be improved by adopting some
local search methods or integration of different strategies for generating initial population. Moreover, the performance of
BBO algorithm can be explored for high-dimension real-world problem which is a challenging task for any algorithm.
References
Abbas, A. T., M. F. Aly, and K. Hamza. 2010. Optimum Drilling Path Planning for a Rectangular Matrix of Holes Using Ant Colony Optimisation. International Journal of Production Research 49 (19): 58775891.
Attar, S. F., M. Mohammadi, and R. Tavakkoli-Moghaddam. 2013. Hybrid Flexible Flowshop Scheduling Problem with Unrelated Parallel Machines and Limited Waiting times. The International Journal of Advanced Manufacturing Technology 68 (58): 117.
Dawei, D., D. Simon, and M. Ergezer. 2009. Biogeography-based Optimization Combined with Evolutionary Strategy and Immigration Refusal. In Proceeding of the IEEE Conference on Systems, Man and Cybernetics, October 2009, 9971002. San Antonio,
TX.
Ghaiebi, H., and M. Solimanpur. 2007. An Ant Algorithm for Optimization of Hole-making Operations. Computers & Industrial
Engineering 52 (2): 308319.
Jamuna, K., and K. S. Swarup. 2012. Multi-objective Biogeography Based Optimization for Optimal PMU Placement. Applied Soft
Computing 12 (5): 15031510.
Kolahan, F., and M. Liang. 2000. Optimization of Hole-making Operations: A Tabu-search Approach. International Journal of
Machine Tools and Manufacture 40 (12): 17351753.
Liu, X., Y. Hong, N. Zhonghua, Q. Jianchang, and Q. Xiaoli. 2013. Process Planning Optimization of Hole-making Operations
Using Ant Colony Algorithm. The International Journal of Advanced Manufacturing Technology 69 (14): 753769.
Ma, H., and D. Simon. 2011. Blended Biogeography-based Optimization for Constrained Optimization. Engineering Applications of
Articial Intelligence 24 (3): 517525.
Medina-Rodrguez, N., O. Montiel-Ross, R. Seplveda, and O. Castillo. 2012. Tool Path Optimization for Computer Numerical Control Machines Based on Parallel Aco. Engineering Letters 20: 101108.
Merchant, M. E. 1985. World Trends and Prospects in Manufacturing Technology. International Journal of Vehicle Design 6 (2):
121138.
Onwubolu, G. C., and M. Clerc. 2004. Optimal Path for Automated Drilling Operations by a New Heuristic Approach Using Particle
Swarm Optimization. International Journal of Production Research 42 (3): 473491.
Oysu, C., and Z. Bingul. 2009. Application of Heuristic and Hybrid-GASA Algorithms to Tool-path Optimization Problem for Minimizing Airtime during Machining. Engineering Applications of Articial Intelligence 22 (3): 389396.
Rahmati, S., and M. Zandieh. 2012. A New Biogeography-based Optimization (BBO) Algorithm for the Flexible Job Shop Scheduling Problem. The International Journal of Advanced Manufacturing Technology 58 (912): 11151129.
Simon, D. 2008. Biogeography-based Optimization. IEEE Transactions on Evolutionary Computation 12 (6): 702713.
Zhang, C., Y. Rao, P. Li, and X. Shao. 2007. Bilevel Genetic Algorithm for the Flexible Job-shop Scheduling Problem. Chinese
Journal of Mechanical Engineering 43 (4): 119124.